SEISMIC HAZARD ASSESSMENT AND SITE-DEPENDENT RESPONSE SPECTRA PARAMETERS OF THE CURRENT SEISMIC DESIGN CODE IN ALBANIA

Transcription

1 SEISMIC HAZARD ASSESSMENT AND SITE-DEPENDENT RESPONSE SPECTRA PARAMETERS OF THE CURRENT SEISMIC DESIGN CODE IN ALBANIA Ll. Duni & N. Kuka Institute of Seismology, Academy of Sciences, Tirana, Albania ABSTRACT: A probabilistically based hazard map of Albania, following the spatially smoothed seismicity approach is presented. The results show the rather great influence of the attenuation models in the results, reaching 25-3% difference in the evaluation of PGA by using Sabetta & Pugliese (1996) and Ambraseys et al. (1996) attenuation models. Taking advantage of this development, an effort is made to analyze the site dependent response spectra parameters of the current seismic design code in Albania. KEYWORDS: Seismic hazard, response spectra, designs code. INTRODUCTION The continuous improvement of procedures for defining the seismic hazard at regional (national) and local level is essential for the optimum design of earthquake-resistant structures. Reference motion and detailed characterization of soil conditions are milestones for the definition of seismic action in design codes. Following the new probabilistically based seismic hazard map of the country recently carried out at the Seismological Institute, and due to the fact that the seismic hazard procedures adopted for the definition of the so-called seismic coefficients of the actual seismic design code are not based on the PSHA (Cornel approach), it is necessary to review the site-dependent spectral response parameters adopted in the current Albanian seismic code KTP-N.2-89, established mostly, taken into account the relations recommended by the MSK-64 macroseismic scale (Koçiu, 23). Furthermore, the new seismic hazard map can serve as basis for the definition of a new seismic design code of the country following the EC8 approach. Seismic hazard analysis is the evaluation of potentially damaging related-phenomena to which a region or a facility may be subjected during its useful lifetime. The simple evaluation of the probable seismic hazard is the mapping of earthquakes epicenters that have occurred historically in a certain area. Another step would be the map of maximum observed intensities. Up to now, the seismic hazard in Albania has been assessed mostly in terms of macroseismic intensity (Sulstarova et al., 198). Several attempts have been made to express the seismic hazard in terms of ground acceleration, velocity and displacement following both deterministic and probabilistic approaches (Muço et al. 21, Muço et al. 22, Peçi et al. 22). The current practice in Albania for assigning the earthquake load in regard to the design of structures is to utilize the seismic zonation map (Fig.1) published by the Seismological Institute in 198 (Sulstarova et al., 198) and various maps compiled during the microzonation studies recently carried out for seven largest urban areas of the country. The map presented in Fig. 1 is based on the intensities of strong historical earthquakes, the earthquakes of XX-th century, as well as on seismotectonic synthesis. According to that, all the territory of Albania is divided into three main zones with basic intensity of shaking VIII, VII and VI degrees of MSK-64 scale, for average soil condition. In some parts, due to poor soil conditions, the seismic intensity may attain up to IX degree of intensity. As average soil condition, thick, stiff quaternary sediments with deep ground water level are considered.

2 SEISMIC HAZARD ASSESSMENT Of the various methods of seismic hazard analysis in use today, the most widespread is the probabilistic seismic hazard analysis (PSHA), proposed by Cornell in 1968 (Cornell, 1968). A PSHA requires a model consisting of three main elements: The definition of potential seismic sources. A statistical description of seismicity in these zones. This is most often expressed in terms of a Guttenberg-Richter (G-R) magnitude recurrence relationship. It is generally assumed that the seismicity of each source follows a Poissonian occurrence process. An attenuation function relevant to the hazard parameters considered. It is necessary to know the rate at which ground motion decays with distance from the epicenter or hypocenter or fault rupture as a function of magnitude. The PSHA output is defined as the probability that a ground motion parameter (PGA, spectral values, intensity, etc.) will be exceeded within a given time period. The result can be expressed either as a hazard curve giving the annual probability of any level of ground shaking being exceeded at the site of interest or in the form of hazard maps representing spatial variability of the selected ground motion parameter for a given return period. Two potential difficulties one faces in the process of probabilistic seismic hazard analysis: the incompleteness of seismic catalogues, Figure 1. Seismic zonation map of Albania, (Sulstarova et al., 198) requirement of specifying seismic source zones. The latest requires many times the expertise of a number of independent groups of specialists for the creation of different, more or less subjective source zone models. On the other hand, the assumption of spatial uniformity within an area source sometimes conflicts with the observed spatial distribution of epicenters. This is the reason why a zoneless approach is often used (Frankel, 1995; Woo, 1996; Crespo et al., 22; Martin et al., 22). Among different methods proposed, we choused the spatially smoothed seismicity approach, developed by Frankel (1995) and extended by Lapajne et al., (1997) to include the seismotectonics characteristics. The method still follows the basic approach of Cornell, but no delineation of seismic sources is needed. The observed area is divided into grid cells, and in each cell the activity rate is calculated and then spatially smoothed. Smoothing is carried out in two stages: the first, a two-dimensional Gaussian smoothing due to inaccuracy of the epicenters location; and the second, a fault rupture-oriented elliptical smoothing, according to the orientations of seismogenic faults in different tectonic regions. The radius of circular Gaussian smoothing is defined in a more objective way from the error of epicenter location and from the presupposed subsurface fault rupture length corresponding to the upper bound magnitude. The annual rate of exceedance of the specified values for a given strong motion parameter, and finally its value for a given return period are calculated. The incorporation of seismotectonic data allows the use of ground motion models based on the shortest distance to fault ruptures and not only to epicenters. The influence of large historical events on the seismic hazard is enhanced

3 LEGEND 4.5 to to to to to to 7.5 Spatial distribution of the epicenters (Time span: 58-2, Ms>=4.5) Figure 2. Map of earthquakes epicenters. Time span 58-2, Ms by adding seismicity models based on the seismic energy released. The adopted approach considers different alternatives about fundamental hypothesis on input parameters to account for and to propagate uncertainties in the model within a logic-tree framework. As seismological database, we used a homogeneous catalogue, compiled by Sulstarova et al., (22), which contains 53 earthquakes with magnitudes greater or equal to 4.5. It covers a time span of 1943 years (58-2), and an area between E and N. The size of the earthquakes is given in terms of surface-wave magnitude M s. A map which depicts the spatial distribution of epicenters of earthquakes used in the present study is demonstrated in Fig. 2. It shows that seismicity is not uniformly distributed within the country. After investigating the completeness of the earthquake catalogue and declustering it from aftershocks and foreshocks, the parameters of the magnitude-frequency relation were estimated by the maximum likelihood method, considering only the independent events of the complete part of the catalogue (Sulstarova et al., 23; Kuka et al., 23). Then, the double-truncated exponential recurrence relationships with b value equal to.9 and M max =7.2 is further used in the seismic hazard computations. To include the seismotectonics characteristics in the hazard computations, a seismotectonic file has been prepared for the territory of Albania which tents to describe the seismotectonic information in a quantitative way. This seismotectonic 43. model is shown in Fig. 3 (Aliaj, 22). Every zone is A A5 characterized by one or more predominant tectonic structures, their orientation and corresponding weights (the 42.4 A1.12 sum of all weights must equal 1), obtained through a A7 statistical analysis of seismogenic zones The seismic hazard was calculated over a grid of A in longitude and.83 in latitude (about x1km), for a total number of 12 computation nodes, 41.4 A2 which cover the entire observed area ( E, N). In each cell of this grid, the number of 41. earthquakes with magnitude equal or greater than the lower 4.8 A9 bound magnitude M min is counted. A8 4.6 Since historical data records are incomplete, it is important that the probabilistic model accounts for this A deficiency. The definition and application of several completeness intervals, the use of different observation 39.8 periods and magnitude ranges gives an estimate of the 39.6 variations of the modeled parameters, which is important 39.4 for successful hazard estimation. Thus, five alternative 39.2 models of the spatial seismic activity were investigated because it was believed that the accuracy and certainty of Figure 3. Quantitative seismotectonic data related to more recent periods is higher, while a model of Albania (Aliaj, 22) comparison of different models may be relevant. In the first two models, the activity rate is counted while foreshocks and aftershocks were removed from the calculations in order to follow a

4 Poissonian earthquake process. Model M1c is based on the complete part of the catalogue, The maximum epicenter location error for this period is estimated about 3 km, so the corresponding smoothing correlation distance c=1km is used. Model M2c is based on large earthquakes with Ms 6. for the period There are 76 such events and the maximum observed magnitude is 7.2. As for the smoothing correlation distance, we used c=12.5km. Due to the requirement that sub-catalogues should be complete, no earthquakes before 155 have been considered yet. However, to emphasize earthquakes with large seismic energy, the alternative seismic activity rate is calculated from released seismic energy from earthquakes including foreshocks and aftershocks (Lapajne, 2). Thus, three additional models M1e, M2e, and M3e are also investigated. Models M1e and M2e are based on the same data as models M1c and M2c, respectively. Model M3e use the whole earthquake catalogue for the period 58-2, with a correlation distance c=15km. The upper bound magnitude M max for the all models was set to 7.2, because this is the highest magnitude observed historically. On the other hand, the maximum observed magnitude since 192 is 6.9, so the value 7.2 for the maximum expected magnitude seems to be reliable, keeping in mind the long return periods of the large events. All models of seismic activity were then smoothed in a two-stage procedure. The fault rupture-oriented smoothing is based on the seismotectonic model of Albania (Fig. 3). They were also normalized, so that the total activity rate in the observed area to be the same in all models. Normalizing of the activity rate of each model to M1c model is performed multiplying it by the ratio of the expected and actual annual activity rates. The main characteristics of the five spatial seismic activity models are shown in Table 1. Table 1. Main characteristics of the spatial seismic activity models Model M 1C M 2C M 1E M 2E M 3E Number of events with M S M min Span period (years) Maximum location error (km) Correlation distance Decay rate, b Lower bound magnitude M min Upper bound magnitude M max Normalization factor Weight of the model The PGA hazard maps have been computed for a 1% exceedance probability in 5 years, corresponding to a return period of 475 years. This is a standard practice in seismic design. All the calculations have been performed making it use the computer code OHAZ (Zabukovec et al., 2). There is not, in fact, a single parameter, which adequately represents the full information of seismic hazard. The most popular parameter is PGA but it is generally associated with a short impulse of very high frequency and, therefore, cannot be easily correlated to the damage observed. For these reasons, spectral accelerations and uniform hazard response spectra have been considered as well. For the five models, the ground motion parameters were calculated using the strong ground motion attenuation models of Sabetta & Pugliese (1996) and Ambraseys et al., (1996), for the larger horizontal component and fault distance. The conventional results of the seismic hazard evaluation for PGA are presented in the corresponding maps. All these maps, which refer to rock condition, have been computed for a return period T=475 years, corresponding to the 9% non-exceedance probability in 5 years. The weighted mean map of the five models by Sabetta & Pugliese (1996) attenuation model is shown in Fig. 4. The same map calculated using the Ambraseys et al., (1996) attenuation model and the worst-case scenario map, showing the highest PGA value from the five models at each location, were also computed.

5 ( ) Figure 4. Map of PGA, 475 years of RP. Sabetta & Pugliese (1996)- stiff soil (mean of 5 models). Comparing results, although the same configuration is maintained, a significant difference is observed, with the Ambraseys et al., (1996) attenuation model giving higher values of about 25-3% than Sabetta & Pugliese (1996) one, for annual probability of exceedance equal to.21 (475 years return period). Considering the most favorable scenario of seismic hazard for Albania, presented in Fig. 4 (derived from the Sabetta & Pugliese (1996) attenuation model) we can say that very few areas of the country can be considered safe, part of Northern Albania, where values less than.15g are expected. The northwestern and the southern part of the country represent the area with the highest hazard according to all the elaborations carried out. Lushnja-Elbasan-Diber area also reveals a high hazard. Acceleration ranges from.2g approximately in all the territory, up to.3g in northwestern and southwestern part of the country. Finally, comparing the maps of Fig. 4 with the actual in force map of seismic zonation of Albania presented in Fig. 1, it is easy to deduce that these two maps have approximatively the same hazard configuration. REPRESENTATION OF SEISMIC ACTION IN THE CURRENT SEISMIC DESIGN CODE The seismic action in the KTP-N.2-89 design code is expressed by an elastic ground acceleration response spectrum Sa(T) defined by the relation: Sa(T) = k E β g (1) where k E is the so-called seismic coefficient, β(t) is the dynamic coefficient (depending on the vibration period T) having the shape shown in Fig. 5 and g is the acceleration gravity. Introducing the coefficients k r (building importance coefficient) and ψ (ductility and damping structure s coefficient), the design acceleration values are obtained. Both k E and β(t) are dependent on local soil conditions, classified in three categories, according to the Table 2. Table 2. Global description of subsoil conditions within the scope of KTP-N.2-89 Soil Category Lithological description of subsoil conditions I II III I.a. Hard rocks, magmatic, partly metamorphized and sedimentary rocks with high static and dynamic stability, I.b. Average strength flysch formations not influenced by tectonic or alteration phenomena, sand stones, conglomerate etc. II.a. Rock formations with developed jointing and alteration phenomena, II.b. Stiff or semi-stiff silty clay formations independently of water content, II.c. Loose formations: 1. Sandy and silty clays, clays in the strong plastic and elastic state with saturation, 2. Stiff or semi-stiff sands and gravels with saturation. III. Loose formations: 1.Large, middle and small grain size sands, dusty sands with near surface water level, 2. Clays and soft up to flowing state plastic silty clays.

6 The values for the seismic coefficient k E are shown in Table 3, while the values that determine the shape of the dynamic coefficient β(t) curves are shown in Table 4. By definition, in KTP- N.2-89 the product k E g is the peak ground acceleration. Table 3. Values of seismic coefficient k E Soil category Seismic Intensity (MSK-64) VII VIII IX I II III Table 4. Values of various parameters defining the spectral shape of β(t) curves Soil category T C(sec) T D(sec) β( T T C ) β(t C T T D) β(t D T) I /T.65 II /T.65 III /T.65 Dynamic coefficient Category I Category II Category III Figure 5 Dynamic coefficient β(t) curves From Table 2 it can be noticed that in KTP-N.2-89, site conditions are specified only on the basis of lithological description of subsoil profile, without taking into account any other parameter, like, for example, shear wave velocity. On the bases of parameters given in Tables 3, 4 and expression (1), site dependent acceleration response spectra for three different levels of seismic intensity can be constructed (Fig. 6-8). From the Fig. 6-8 it can easily understood that T C and T D represent the end of the upper constant portion of spectral acceleration and the beginning of the lower constant portion of spectral acceleration, respectively. DISCUSSION OF RESPONSE SPECTRA PARAMETERS Seismic hazard assessment and assessment of the seismic design criteria for conventional structures is a very important task. It is already a common practice in seismically active areas to assess the seismic hazard and design criteria using the probabilistic approach, which permits to add in the analysis the quantitative notion of earthquake return period. This helps the engineers to fix the reference design seismic motion for this kind of buildings (and not only) referring to specific return period. Taking advantage of hazard map shown in Fig. 4 it is possible to analyze the level of protection that deterministically based seismic coefficients of KTP-N.2-89 offer in comparison with the values of PGA determined through probabilistic procedures. For simplicity we analyze first the seismic coefficients for category I (rock and firm soil conditions) for three different intensity levels according to the MSK-64 scale, respectively VII, VIII and IX degrees. We consider as representative areas: Tirana for intensity VII, Elbasani for intensity VIII and Vlora and Pogradeci for intensity IX. For Tirana area our evaluations give the value of peak ground acceleration.2g for rock conditions and 475 years of return period (Duni, 23). The applied value of peak ground acceleration for this area according KTP-N.2-89 is.8g, which corresponds to a return period of 5 years, according to the methodology described previously.

7 Sa (cm /s/s) Category I Category II Category III Sa(cm/s/s) Category I Category II Category III Sa(cm/s/s) Category I Category II Category III Figure 6. Site-dependent response spectra for intensity VII Figure 7. Site-dependent response spectra for intensity VIII Figure 8. Site-dependent response spectra for intensity IX For Elbasani area, the map of Fig. 4 gives a value of.25g for the peak ground acceleration, while the KTP-N.2-89 foresees a value of.16g, which has a return period of 15 years. For the areas of the country with expected intensity IX degree, we have values of peak ground acceleration.27g and.22g for Vlora and Pogradeci towns, respectively. For Vlora area the KTP-N.2-89 prevents the same value of.27g having the same return period of 475 years, while for Pogradeci the KTP-N.2-89 overestimates the value of peak ground acceleration, which has in this case, a return period higher than 475 years. If we take into account the values of peak ground acceleration assessed following the probabilistic approach for soil conditions (using appropriate strong motion attenuation models) and compare them with the seismic coefficients of KTP-N.2-89 for soil category II and seismic intensity VII, taking as representative the up to now investigations for the area of Tirana, we observe the same trend as those of soil category I i.e., the return period for the seismic coefficient k E =.11 is 5 years (Duni, 23). Another aspect of interest regarding the site-dependent response spectra parameters of the actually in force design code is the amplification level that seismic coefficients offer for different seismic intensity zones and soil conditions. In the absence of representative strong motion records at various intensity levels in Albania, comparisons can be made with variation of PGA proposed for the European region by EC8 (EC8, 1996). Taking into account the peak ground acceleration values shown in Table 3 and accepting the values for soil category I as reference ones, the relations a s /a r can be expressed in the way shown in Table 5, where a r and a s are peak ground acceleration on rock and soil sites, respectively. From the values of relation a s /a r shown in Table 5 it can be noticed that KTP-N.2-89 predicts larger values of peak ground acceleration for weaker soils in comparison to the reference one, especially for the soil category III. The values of Table 5 are higher than respective values offered by the EC8 (EC8, 1996). Table 5. Values of relation a s/a r according to KTP-N.2-89 Soil category Seismic Intensity (MSK-64) VII VIII IX I II III Let s finally discuss the shape of spectra proposed by the KTP-N.2-89 design code and uniform hazard spectra constructed for an annual probability of exceedance of.21, i.e. for 475 years of return period for the four towns mentioned above, determined using the PSHA described previously. The probabilistic spectra have been calculated for five periods of engineering interest.2,.3,.5, 1. and 2. sec and 5 % of the critical damping. The strong motion attenuation model of Ambraseys et al., (1996) has been used and the uncertainty of the attenuation model has been taken into account by considering the standard deviation in the computations. In Fig. 9 presented are the 5% damping elastic response spectra for the area of Tirana City, respectively the spectra for the category II of soil that prevail in Tirana and seismic intensity VII which belongs the area according to the KTP-N.2-89 (Sulstarova et al., 198), the elastic

8 response spectra according the EC8 for soil class B defined according the parameters given in the appropriate table of EC8 (EC8, 1996) accepting as input motion intensity the value of a g =.8g that corresponds to the value of k E g for the soil category I of KTP-N.2-89 (that in this case is considered equal to the rock or firm soil of class A of EC8) and the uniform hazard spectra. We can observe that there is a big difference in the level of spectral response in all the periods of engineering interest predicted by the uniform hazard spectra and spectra according the KTP-N.2-89 currently in use in Albania for this area. Considering the values of Sa for T=1sec, that is approximatively the vibration period of high rise buildings that are very rapidly spreading all around the territory of the town, the relation Sa UHS /Sa KTP-N.2-89 is 3, while the relation Sa UHS /Sa EC8 is 2.2. That means that although the EC8 response spectra has been scaled with the value of peak ground acceleration predicted by the actual design code, it presents a better approach than that actually in use for the area of Tirana. In Fig. 1 the same spectra as those presented in Fig.9 for the Elbasani town are shown Sa (cm/s/s) Category II, ao=.8g Class B, EC8 UHS for Tirana Figure 9. Comparison of response spectra for the area of Tirana Sa (cm/s/s) Category II, ao=.16g Class B, EC8 UHS for Elbasani Figure 1. Comparison of response spectra for the area of Elbasani with the difference that the elastic response spectra according the KTP-N.2-89 represent that of category II of soil and seismic intensity VIII to which the area of Elbasani belongs, while the elastic spectra of EC8 for the subsoil class B has been scaled with the value of peak ground acceleration.16g. The same tendency of spectral response is observed as in Tirana case, but the level of disagreement among the three spectra is lower. The relation Sa UHS /Sa KTP-N.2-89 for period T=1sec is 1.8, while the relation Sa UHS /Sa EC8 is 1.3. Here again the EC8 elastic response spectra presents a better approach than that actually in use for the area of Elbasani. In Fig. 11 and 12 presented are the elastic response spectra for the towns of Vlora and Pogradeci, both belonging to the area of seismic intensity IX degree of MSK-64 scale according the seismic zonation map of Albania (Sulstarova et al., 198). In these two cases the spectra according the actual in force code KTP-N.2-89 are those for the category III of soil, while the spectra of EC8 for the subsoil class C, characteristic for these two towns, have been scaled with the value of peak ground acceleration.27g. Here the spectra behavior is different compared to the Tirana and Elbasani cases. For Vlora area the three spectra prevent quite the same level of Sa(cm/s/s) Category III, ao=.27g Class C, EC8 UHS f or Vlora Figure 11. Comparison of response spectra for the area of Vlora Sa(cm/s/s) Category III, ao=.27g Class C, EC8 UHS for Pogradeci Figure 12. Comparison of response spectra for the area of Pogradeci

9 spectral response for a large values of periods of engineering interest, showing that the elastic response spectra according the KTP-N.2-89 for the category III of soil conditions represents a good approach for high rise buildings compared to EC8 one. For Pogradeci area, the elastic response spectra of KTP-N.2-89 and EC8 look even conservative in preventing the spectral ordinates over all the periods of engineering interest compared to the uniform hazard spectra. CONCLUSIONS Taken advantage of the recent improvements in seismic hazard assessment in national scale, an effort has been made in this paper to analyze the site dependent response spectra parameters applied in the current seismic design code KTP-N.2-89 in Albania. The analysis show that the so-called seismic coefficient of this code prevent large peak ground accelerations for softer soils compared with EC8 (EC8, 1996) especially for seismic intensity zone VII. As far as probability levels of these seismic coefficient is concerned, comparisons with recent studies in the framework of application of probabilistic assessment of seismic hazard for site response analyses in various areas of the country (Duni, 23), show that seismic coefficients of KTP- N.2-89 for soil category I and seismic intensity zone VII (PGA=.8g) has a return period of approximatively 5 years for Tirana area; for seismic intensity zone VIII (PGA=.16g) 15 years return period, considering Elbasani area as representative; and for seismic intensity zone IX (PGA=.27g) represented in the analysis by case studies of Vlora and Pogradeci, results show return periods equal or larger than 475 years. These results are derived from the Sabetta & Pugliese (1996) attenuation model. Analysis for soil category II and seismic intensity VII are available only for the area of Tirana and show the same trend as for soil category I, i.e., the return period for the seismic coefficient k E =.11 (PGA=.11g) is 5 years (Duni, 23). The same results are shown also for the spectral response predicted by the site dependent response spectra. The largest difference is observed for the seismic intensity zone VII, while for seismic zone intensity IX the response spectra of KTP-N.2-89 design code predict quite the same level of spectral response as uniform hazard spectra evaluated according the PSHA (spatially smoothed seismicity approach) for 475 years of return period, that is already an accepted standard in design practice in Europe (EC8, 1996). While the authors of this paper have join their efforts for a rapid implementation of EC8 in Albania as soon as possible, they think that a partial modification of the actual macroseismic based seismic coefficients of the actual seismic design code KTP-N.2-89 in Albania is necessary, especially for the seismic intensity VII zone. The increase of this seismic coefficient for this zone will have its influence in the amplitude of response spectra that will reach higher level of spectral response, aiming to follow the response predicted by probabilistically based response spectra. This is necessary because, even in the case of a rapid implementation of EC8 in Albania, it would be necessary the two codes to coexist for some time in order the engineers to be familiar with the philosophy of the new code and to have the possibility to compare the force levels and design procedures predicted by the two above mentioned codes. REFERENCES Aliaj Sh., 22, Seismotectonic model used for hazard calculations of Albania according to OHAZ computer code, Archive of Seismological Institute, Tirana, Albania. Ambraseys N. N., Simpson K. A., Bommer J. J., 1996, Prediction of horizontal response spectra in Europe, Earthquake Engineering and Structural Dynamics, Vol. 25, pp Cornell C. A., 1968, Engineering seismic risk analysis, Bull. Seis. Soc. Am, V. 58, pp Crespo M.J. & Marti J., 22, The use of zoneless method in four LNG sites in Spain, in 12 th European Conference on Earthquake Engineering, September 9-13, 22, Paper Ref. 36, pub. by Elsevier Science Ltd.

10 Duni Ll., 23, The application of strong motion attenuation models for the assessment of seismic hazard of construction sites in terms of strong motion parameters, Internal study, Archive of Seismological Institute, Tirana, Albania, pp. 56, (in Albanian). Eurocode 8, 1996, Design provisions for earthquake resistance of structures, DD ENV Frankel A., 1995, Mapping Seismic Hazard in the Central and Eastern United States, Seismological Research Letters, V. 66, N. 4, July-August 1995, pp Koçiu S., 23, Relation on determination of ground strong motion shaking in construction sites, Internal report, Archive of Seismological Institute, Tirana, Albania, pp. 26, (in Albanian). KTP-N.2-89, 1989, Earthquake resistant design provisions, Publication of Academy of Sciences and Ministry of Constructions, Tirana, Albania, pp. 17, (in Albanian). Kuka N., Sulstarova E., Duni Ll., Aliaj Sh., 23, Seismic hazard assessment of Albania by spatially smoothed seismicity approach, in Book of abstracts of SE 4EEE, Skopje-Ohrid, August 23, pp. 8. Lapajne J., Sket-Motnikar B., Zabukovec B., Zupancic P., 1997, Preliminary Seismic Hazard maps of Slovenia. Natural Hazards, Vol. I, 14, Nos.2-3, pp Lapajne J. K., 2, Some Features of the Spatially Smoothed Seismicity, Proceedings of the workshop on Seismicity Modelling in Seismic Hazard Mapping, Poljce, Slovenia, May 22 24, 2, pp. 7. Martin C., Secanell R., Combes Ph., Lignon G., 22, Preliminary Probabilistic Seismic Hazard Assessment of France, in 12 th European Conference on Earthquake Engineering, September 9-13, 22, Paper Ref. 87, pub. By Elsevier Science Ltd. Muço B., Vaccari F., Panza G., 21, Seismic zonation of Albania using a deterministic approach, Albanian Journal of Natural & Technical Sciences, Published by Academy of Sciences of Albania, 21 (1), pp Muço B., Kiratzi A., Aliaj Sh., Sulstarova E., Kociu S., Peçi V., 22, Probabilistic Approach Used in Seismic Hazard of Albania, Book of Abstracts, Univ. of Genoa, Italy, 28-th General Ass. of ESC, September 1-6, 22. Peci V., Adams J., Al-Khoubbi I., Halchuk S., 22, Probabilistic Seismic Hazard Maps for Albania, Book of Abstracts, Univ. of Genoa, Italy, 28-th General Ass. of ESC, September 1-6, 22. Sabetta F., Pugliese A., 1996, Estimation of response spectra and simulation of nonstationary earthquake ground motions, Bull. Seism. Soc Am., Vol. 86 No. 2, pp Sulstarova E., Koçiu S., Aliaj Sh., 198, Seismic zonation of Albania, Publication of Seismological Center of Academy of Sciences, Tirana, Albania, pp Sulstarova E., Koçiu S., Muço B., Peçi V, Duni Ll., 22, Catalogue of Strong Albanian Earthquakes from 58 to 2, Internal report, Archive of Seismological Institute, Tirana, Albania. Sulstarova E., Aliaj Sh., Muço B., Kuka N., Duni Ll., 23, Earthquakes, seismic hazard and seismic risk in Albania, Technical report, in the framework of UNDP project Natural risks of Albania, Archive of Seismological Institute, Tirana, Albania, pp. 97. Zabukovec B., Motnikar B. S., Zupancic P., 2, OHAZ: A Computer Program for Seismic Hazard Calculation, User Manual, Geophysical Survey of Slovenia, pp. 48. Woo G., 1996, Kernel Estimation Methods for Seismic Hazard Area Source Modeling, Bull. Seism. Soc Am., Vol. 86, No. 2, pp