Seismic slope stability assessment of a construction site at Bandırma, Turkey

Transcription

1 Seismic slope stability assessment of a construction site at Bandırma, Turkey Serkan Üçer Research Assistant, Civil Engineering Department, Middle East Technical University, Ankara, Turkey userkan@metu.edu.tr Selman Sağlam Research Assistant, Civil Engineering Department, Middle East Technical University, Ankara, Turkey selman@metu.edu.tr Sadık Bakır Associate Professor, Civil Engineering Department, Middle East Technical University, Ankara, Turkey bakir@metu.edu.tr KEYWORDS: Seismic slope stability, permanent displacement ABSTRACT: Seismic slope stability assessment for a power plant construction site at Bandırma, Turkey is presented. The site, which was evidently subjected to earlier slope failures, was analyzed using limit equilibrium and stress deformation methods. Subsurface characteristics were determined through investigations including boreholes reaching 100 m depths accompanied with penetration testing (SPT and CPT), soil sampling, laboratory testing and detailed geophysical surveys. Conventional slope stability analyses were utilized to identify the most critical slide surfaces for the investigated site, and the seismic slope stability is analyzed through pseudo static and dynamic methods. The seismic coefficients and acceleration time histories used in the analyses were determined considering the seismic characteristics of the site. Performance of the slopes evaluated from both methods is discussed in a comparative manner. Finally, based on the results of the analyses, possible countermeasures against the landslide hazard were investigated from the point of effectiveness and cost. 1 INTRODUCTION Slope failures are one of the primary causes of life and property loss during earthquakes. In this paper, seismic performance of a thermal power plant construction site situated over an ancient landslide mass is evaluated. The construction site in question is located at the shoreline near the town Bandırma in the Region of Marmara in Turkey. Considering that the Region of Marmara has a rather high potential of seismic activity in general, and that the site of interest is particularly close to the southern segments of the North Anatolian Fault System in the area, the assessment of the landslide hazard at the site is of great importance. The performance of the site consisting of slopes of variable inclinations was first investigated via conventional slope stability analyses to identify the most critical slide surfaces. Then, seismic stability of the critical slide surfaces were analyzed with pseudo-static and dynamic approaches. Additionally, the methodology suggested by Makdisi and Seed (1978) was utilized to estimate permanent displacements. Consistent with the seismicity of the region, dynamic analyses were performed considering both peak ground accelerations (for pseudo-static analyses) and actual earthquake time histories (for dynamic finite element analyses). Safety factors evaluated from both 290

2 static and seismic slope stability methods are discussed in a comparative manner. Finally, based on the results of analyses, possible countermeasures against the landslide hazard prevention were investigated from the point of effectiveness. 2 OVERVIEW OF SEISMIC SLOPE STABILITY ASSESSMENT METHODS In the pseudo-static approach, the earthquake effects on a potential sliding mass is represented by horizontal and vertical seismic coefficients. The inertial forces produced by an earthquake shaking are expressed as the product of seismic coefficients and the weight of potential failure mass. Accordingly, the factor of safety is expressed as the ratio of the resisting forces to the driving forces. The results of pseudo-static analyses critically depend on the horizontal seismic coefficient (k h ) rather than that of the vertical, which is frequently ignored. Therefore, selection of an appropriate seismic coefficient is the most important aspect of a pseudo-static analysis. Terzaghi (1950) suggested the use of k h values of 0.1, 0.2 and 0.5g depending on the earthquake severity. Seed (1979) suggested that the safety factors become greater than unity for the slopes of earth dams if the seismic coefficients are taken as between 0.10 and It should be noted that the soil materials used in construction of earth dams do not show more than 15% strength loss upon cyclic loading. Marcuson (1981) recommended using one-third or one-half of the maximum acceleration as seismic coefficients for dams. Hynes-Griffin and Franklin (1984) recommended that use of seismic coefficient (k h ) equal to 50% of the peak ground acceleration would not develop large deformations in the earth dams with pseudo-static factors of safety greater than unity. Although there exist various recommendations on the selection of seismic coefficients, it seems clear that the pseudo-static coefficient is fairly related with the anticipated or experienced peak ground acceleration and the engineering judgment. Seismically induced permanent deformations of slopes are commonly estimated by means of the procedures based on the landmark work of Newmark (1965), in which the potential landslide is represented by a rigid block resting on an inclined plane, and the expected displacement is calculated by integration of the equation of motion for the rigid block. Makdisi and Seed (1978) proposed a simplified approach based on the sliding block model to compute earthquake induced permanent displacements of earth dams and embankments. The yield acceleration (a y ) for a particular potential failure surface is computed using the dynamic yield strength which was proposed as 80% of the undrained strength of the dam material. The permanent displacement is then computed by means of the relationship representing the variation of permanent displacement with the ratio of yield acceleration to average maximum acceleration (a y /a max ) and the earthquake magnitude. Stress-deformation analyses of seismic slope stability are often carried out through dynamic finite element method programs. The seismically induced total permanent deformation of a slope is calculated through integrating the seismically induced permanent deformations occurring instantaneously during shaking. 3 INVESTIGATION OF THE SITE The property, over a part of which the thermal power plant is planned to be built, is located on the southern coast of the Sea of Marmara, at Bandırma. The land has about 700 m shoreline and extends approximately 3.5 km to the south. Rolling hills dominate the terrain with heights reaching in excess of 190 m above the sea level within the property. Natural slopes existing in the area are around 20 o to 25 o. The investigation aiming at evaluation of the proposed site for the thermal power plant consisted of geophysical and geotechnical investigations phases, each of which is briefly summarized below. 3.1 Geophysical Investigations 291

3 Seismic Slope Stability Assessment of a Construction Site at Bandırma, Turkey Üçer S., Sağlam S., Bakır S Geophysical investigations revealed existence of faults closer to the shoreline within a distance of about 650 m. One of those apparently sheared the recent sea bed sedimentation, indicating recent activity. From a point of consideration of hazard to the site in question, however, none of the alignments of these faults, either active or dormant, crosses the site of concern. An unexpected finding based on the results of geophysical explorations was the nonexistence of a formation that could be interpreted as bedrock neither onshore nor offshore within a depth of about 100 m from the surface. This surprising outcome, which was confirmed later at least in part through the boreholes extending to such depths can possibly be attributed to the existence of a paleo fault (i.e., an ancient fault which led to shearing and displacement of part of the bedrock formation in due course). Although the seismic site classification can be assessed based on other parameters as well, the measured shear wave velocity (V s ) profile is generally considered to be more reliable and preferred for this purpose. Variation of the measured shear wave velocity at the site was as follows: 0-5 m 200 m/s 5-10 m 250 m/s m 300 m/s >25 m 400 m/s The variations detected in measured properties during geophysical surveys within the ranges of depth intervals between 20 m 30 m and between 50 m 70 m could be interpreted as possible zones through which past failures occurred. On the contrary, however, only insignificant differences were recorded in electrical resistivity along the investigated depths, indicating existence of a uniform lithological character in general. Other geophysical measurements also supported this finding. 3.2 Geotechnical Investigations Total of 49 boreholes, 7 of which were offshore, were drilled in the property for the purpose of detailed geotechnical investigation. The emphasis was on the issues of slope stability assessment and foundation design at the power plant site. Borehole depths on the land were by and large around 40 m, except two 100 m and one 80 m boreholes were drilled with sampling of soil cores to obtain information on lithology at greater depths and possible identification of the former slope failure zones. Offshore borehole depths varied between 30 m and 50 m, measured from the sea level. Field tests were performed and disturbed as well as undisturbed samples were retrieved from depths up to 50 m for soil classification, strength and compressibility testing. Consistent with the predictions put forward following geophysical investigations, no bedrock formation was encountered in any of the drillings. Also, based on the inspection of soil core samples, a possible former failure surface was identified at a depth of approximately 30 m, again in conformance with the geophysical investigations and site reconnaissance surveys. Standard classification tests including sieve analyses and Atterberg limits tests were performed in the laboratory on disturbed samples, and soils were classified according to the Unified Soil Classification System (USCS). In general, the silty, sandy and gravelly clay with varying proportions dominated all boreholes beginning to end. This formation, which was classified as low or high plastic clay (CL, CH) consisted of occasional boulder size rocks and sporadic layers of granular soils generally defined as silty, clayey sand and gravel (GC, GW, SC and SM). No ground water of any sort (static or flowing) was encountered in the boreholes. Standard Penetration Test (SPT) blows varied between 20 and refusal, indicated stiff/dense formations. The general tendency was increase in blow counts with increasing depth. Due to the presence of gravel and larger size particles in clay matrix, the SPT test results indicating refusal were deemed to be questionable and were disregarded in evaluations involving SPT data. 292

4 4 ASSESSMENT OF SHEAR STRENGTH PARAMETERS Mean values of the cohesion and internal friction angle parameters measured during triaxial tests can be practically assumed as c = 150 kpa and = 10º. However, this set of shear strength parameters is presumed to consist of a lower bound to the existing range due to two basic reasons: first, the undisturbed samples could only be retrieved from softest/weakest layers of clay (with SPT blow counts at most up to about 25); and second, the level of disturbance during sampling increases with increasing soil hardness, which in turn results in considerable reductions in the laboratory measured strength. Also, it is to be underlined here that a great majority of the samples were retrieved from depths within the top 20 m, since clay was stiffer in general with increasing depth. On the other hand, the uniaxial compression tests yield an average compressive strength of q u = 270 kpa. However, the uniaxial test results are questionable since the samples were unsaturated. Excluding the refusals from SPT data, to be on the safe side, the representative average blow counts for the top 20 m and below, are 35 and 60, respectively. Although, the SPT test is particularly suitable for sands and not recommended for use in strength correlations for cohesive soils, the undrained mass shear strength of clays can be conservatively estimated based on SPT blow counts and plasticity index (Stroud, 1975). Accordingly, for the representative average plasticity index of 18, shear strength parameters of c u = 210 kpa and c u = 360 kpa are calculated. Pressuremeter tests were performed in four of the boreholes. The average values of the undrained cohesion (c u ) are calculated for each of these boreholes are 440 kpa, 580 kpa, 400 kpa and 420 kpa. The results which were found to be questionable, possibly due to partial collapse of the borehole, were eliminated during averaging. Based on an overall evaluation of the laboratory and field test results, the soil profile beneath the site considered for the construction of power plant is idealized into two sub layers with the following conservatively assigned shear strength parameters: Upper Layer, 0 < depth <50 m: c = 250 kpa; = 10º Lower Layer, 50 m < depth: c = 450 kpa; = 10º The unit weight was presumed to be 20 kn/m 3 for both layers based on the laboratory test results. There is no field or laboratory test data relating to soil strength below 50 m due to the limitations involved in in-situ testing or sampling. However, the soil strength typically increases under increasing confining (in-situ) pressures and this general trend has been observed up to 50 m at the site under investigation. 5 SEISMICITY OF THE SITE The seismicity and seismotectonics of the subject area were studied in detail by Yılmaz and Akkar (2008). The site-specific design basis response spectra as well as the peak ground accelerations (PGA) were calculated through probabilistic seismic hazard analysis (PSHA). The hazard was identified for an event with a return period of (T R ) of 2500 years, which is accepted as the definition of maximum considered earthquake (MCE) for the seismically active regions. The spectra are constructed as required by the Turkish Earthquake Code as well as Eurocode 8 and presented for different site classes at the locality of concern. Considering that the site in question is classified as soft with reference to the measured shear wave velocities, the peak ground accelerations were assigned as 0.7g and 0.9g for earthquakes with return periods of 1000 years and 2500 years, respectively, for such sites. The reported site-specific design spectra constructed according to Eurocode 8 (CEN, 2004) and Turkish seismic design provisions (TEC, 2007) are presented in Figure 1a. Five actual earthquake records having response spectra consistent with the site specific design spectra for rock are selected as scenario events for the dynamic analyses. Response spectra and relevant characteristics of these earthquakes are presented in Figure 1b and Table 1, respectively. Since dynamic excitation is 293

5 Seismic Slope Stability Assessment of a Construction Site at Bandırma, Turkey Üçer S., Sağlam S., Bakır S presumed to act at the base of the model which is considered to be the bedrock, records obtained in rock or stiff soil site conditions are preferred. Damping ratio of 5 % is used in evaluation for all of the response spectra. All spectra were normalized with peak ground acceleration. response spectra for selected EQ's response spectra for selected EQ's spectral acc./peak ground acc period (sec) (a) Turkish EQ Code 2007 Eurocode 8 Prob. Seismic Hazard Ass. spectral acc./peak ground acc Earthquake Date distance (km) M PGA (g) PGV (cm/s) PGD (cm) Site Condition Düzce, TURKEY <Vs<360 Landers, USA <Vs Supersitition Hills, USA <Vs<750 Morgan Hill, USA <Vs El Salvador, EL SALVADOR <Vs period (sec) (b) Eurocode 8 Düzce EQ Supersitition Hills EQ Morgan Hill EQ El Salvador EQ Figure 1: (a) Site specific design spectra according to Eurocode 8 and Turkish seismic design provisions, (by Yılmaz and Akkar, 2008) (b) comparison of response spectra of the selected earthquakes with the site specific spectra obtained by Probabilistic Seismic Hazard Assessment As it is seen from Table 1, the original record of Düzce Earthquake is taken on a site of soft soil, so the record was first scaled to 0.94g, then deconvolved to bedrock considering an overlaying soil layer of 75 m before using in the relevant analyses. Table 1 Characteristics of the earthquakes Landers EQ 6 SLOPE STABILITY ASSESMENT 5.1 Background Since the initial configuration of the soil mass before occurrence of the past slide is unknown, a back-analysis type of an approach is not possible for the case in hand. Analyses were conducted using finite element method, limit equilibrium method and Makdisi & Seed (1978) approach on a typical north-south aligned vertical section considering the existing and modified geometries (Figure 2). The strength parameters discussed previously are utilized to model the relevant materials in the analyses (Table 2). The elasticity moduli of the materials used in the dynamic finite element analyses were determined based on the shear wave velocities assigned to the layers The studied section consists of three geometrically distinct parts: the slope along the shoreline (to the north), the slope of the rising hill (to the south) and the almost flat ground of about 250 m length in between the two slopes. The near-flat zone in the middle has an average elevation of 40 m from the sea level. This zone is planned to be leveled at 30 m before the initiation of construction. The slope to the north is variable between almost vertical in some parts to less than 20º. This situation must be due to the ongoing coastal erosion of the landslide mass, which apparently had flown in part into the sea. In the analyses it was utilized a 26º inclination from horizontal for the northern slope. The slope of the rising hill to the south, on the other hand, is approximately uniform at 22º and slightly concave in plan. The naturally existing slope heights for the northern and 294

6 southern slopes are 30 m and 80 m, respectively. The southern slope, in fact, continues to rise further to the south beyond 80 m, with a slightly milder inclination. N Material-3 Sea level Material-4 Construction site Material-2 Material-1 Figure 2 The cross section of the site used in the model Analyses of the section with the naturally existing geometry under static conditions yield safety factors close to 2 (two) for both northern and southern slopes regarding relatively shallow local failure surfaces. Similarly, the factors of safety corresponding to the deep seated shear failures spanning the whole section slightly exceed 2 (two). Considering the rather high seismicity of the area, however, these results are no surprise, and clearly indicate that the natural slopes are controlled by the seismically induced forces in the area. Table 2 The properties of the materials used in the model material no material ' ( 0 ) c ' (kpa) E ' (kpa) V s (m/s) surface clay clay seabed clay bottom clay Stability Analyses, Results and Implications In the analyses, the seismic forces are considered through quasi-static approach, which involves utilization of the static equivalent of dynamic forces applied laterally to the potential sliding mass. The magnitude of the lateral force is expressed via multiplying the weight subject to sliding with a dimensionless pseudo-static acceleration coefficient (k h ), expressed as a fraction of the ratio of lateral acceleration to the gravitational acceleration (g). As mentioned in section 2, the pseudostatic coefficient is generally assigned a value of 1/2 to 1/3 of the ratio a max /g, where a max is the peak ground acceleration. Hence, in view of the peak ground acceleration of 0.9g, specified for the earthquake having 2500 years return period in the area, it was presumed a likely maximum pseudostatic coefficient as 0.4. However, considering that this value is being rather high, it was also utilized coefficients of 0.30 and 0.35 were also utilized. Seismic slope stability analyses were carried out for the section with the ground flattened to 30 m from the sea level. The factor of safety for the northern slope was calculated to be slightly over unity, following flattening even for the pseudo-static coefficient of 0.4 when the slope angle was 26º. This can be attributed to the relatively lower slope height of about 30 m. The factors of safeties calculated from limit equilibrium approach corresponding to the three pseudo-static coefficients, as well as the static condition for the southern slope are tabulated in Table 3. As it can be observed, the static factor of safety of 1.86 reduces slightly below unity for all three pseudo-static coefficients representative of the seismic load conditions, indicating that the stability is likely to be lost momentarily during a major earthquake. The slope movements 295

7 Seismic Slope Stability Assessment of a Construction Site at Bandırma, Turkey Üçer S., Sağlam S., Bakır S corresponding to the factors of safety of below unity were also calculated utilizing the methodology suggested by Makdisi & Seed (1978) and dynamic finite element method. Table 3: Analyses results: (a) Factors of safety corresponding to the static and pseudo-static conditions, and permanent displacements estimated by Makdisi & Seed (1978) method (b) permanent displacements estimated by the dynamic finite element analyses (a) Analysed Conditons Static 0.30g 0.35g 0.40g excavation amount m 3 /m Natural Slope Option 1 Option 2 Option3 FS (cm) FS (cm) FS (cm) FS (cm) (b) Landers Düzce Supersitition Hills Morgan Hill El Salvador PGA (cm) PGA (cm) PGA (cm) PGA (cm) PGA (cm) Furthermore, alternative modifications to the southern slope to improve the seismic response within reasonable economic costs were studied. Calculated safety factors together with the corresponding slope displacements and required amount of excavation per meter thickness of the section are presented in Table 3. The proposed modifications and the respective critical failure surfaces are shown in Figure 3. The seismically induced displacements, remaining within a margin of 10 cm appear to be tolerable in general. However, diverse needs and requirements may require the seismically safer alternative modifications. Figure 3: The modified sections and respective critical failure surfaces Dynamic finite element analyses were performed by PLAXIS V.8 (PLAXIS bv., 2007) software. In these analyses, dynamic excitation was applied from the base of the model as acceleration time histories. Damping was considered in the models by utilizing Rayleigh damping and absorbent model boundaries. Results are presented in Table 3. 5 cm hor. displ. Figure 4: Illustrative horizontal displacement calculation for Supersitition Hills Earthquake 6 CONCLUSIONS 296

8 A thermal power plant construction site situated over an ancient landslide mass is evaluated from the point of the seismic slope stability through pseudo-static method, Makdisi & Seed (1978) approach and dynamic finite element method analyses. The ancient landslide is not currently active, and displays a stable equilibrium under static conditions. However, due to the high seismicity of the area, earthquake induced loads govern the slope performance at the site. Based on an overall evaluation of the laboratory and field test results, the soil profile beneath the site considered for the construction of power plant is idealized into the sub layers with the following conservatively assigned shear strength parameters: c = 250 kpa; = 10º and c = 450 kpa; = 10º. In the analyses, the northern slope is calculated to be stable under the estimated seismic effects, for a slope of 26º. For the southern slope, the safety factors are calculated to be likely to fall below unity during severe earthquakes to be expected in the area. This situation, although implies instability, it should be recognized that occurs only momentarily during an earthquake. The total amount of displacements of the soil mass corresponding to cases of factors of safety of below unity are calculated to be within a margin of 10 cm in the horizontal plane. Alternative modifications to the southern slope to improve seismic response were studied. Safety factors and corresponding slope displacements are presented. The pseudo-static factor of safety (FS) values of modified sections are ranging in a narrow interval about 1 for the k h values of 0.3, 0.35 and 0.4g whereas the static FS values are greater than 2. Although the pseudo- static FS values are near 1, the expected permanent displacements calculated through the method of Makdisi & Seed (1978) are in the level that can be neglected. As it is observed in Table 3, permanent displacements are calculated to range between 2 and 11 centimeters in the dynamic finite element analyses. This finding is quite consistent with the results obtained using the methodology proposed by Makdisi & Seed (1978). Considering the existence of capable faults in the region, however, large scale ground movements, such as those observed during August 12, 1999 Kocaeli earthquake along the southern shore of the İzmit Bay, are also likely to occur during a major earthquake. Since such movements would be driven by the fault activity occurring at great depths, it is not possible to predict their likely nature. However, it is to be emphasized that such a danger exists almost for any site in the area, particularly along the shoreline. REFERENCES CEN. Eurocode 8 (2004). Design of Structures for Earthquake Resistance-Part 1: general rules, seismic actions and rules for buildings, EN :2004. Comite Europeen de Normalisation, Brussels. Hyness-Griffin M.E., Franklin A.G. (1984). Rationalizing the Seismic Coefficient Method. Miscellaneous Paper GL-84-13, US Army Corps of Engineers. 21pp. Makdisi, F. I., Seed, H. B., (1978). Simplified Procedure for Estimating Dam and Embankment Earthquake Induced Deformations. J. of Geotech. Eng. Vol.104, pp Marcuson, W. F. (1981). Moderator s report for session on Earth Dams and Stability of Slopes under Dynamic Loads. Proceedings, International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, St. Louis, Missouri, Vol. 3, p Newmark, N.M. (1965). Effects of Earthquakes on Dams and Embankments. Geotechnique, Vol.5, No.2. PLAXIS bv (2007), PLAXIS: Finite element Package for Analysis of Geotechnical Structures, Delft, Netherland. Seed, H.B. (1979). Considerations in the Earthquake-Resistant Design of Earth and Rockfill Dams. Geotechnique, Vol.29, No.3, pp Stroud, M.A. (1975). Standard Penetration Test in Insensitive Clays and Soft-Rocks. Proc. ESOPTI 2(2). pp Terzaghi, K., (1950). Mechanism of Landslides, The Geological Survey of America, Engineering Geology (Berkeley) Volume. Turkish Earthquake Code (TEC) (2007). Deprem Bölgelerinde Yapılacak Yapılar Hakkında Yönetmelik, General Directorate of Disaster Affairs, Ministry of Public Works and Settlement, Ankara, Turkey Yılmaz, M.T., Akkar, D.S. (2008). Report on: Site Specific Design spectrum and PGA for the Enerjisa A.Ş. Owned Land at Şirinçavuş-Bandırma Using Probabilistic Seismic Hazard Analysis, METU, Ankara, Turkey. 297