Yerel zemin sınıfı ve deprem özelliklerinin tek boyutlu dinamik davranışa etkisi

Abstract

Depremler tekrarlı yüklemelere neden olarak zemin tabakalarının özelliklerini etkilemekte, yeryüzünde deplasmanlara neden olmakta ve üst yapılarda hasara yol açabilmektedir. Üstyapı tasarımlarında kullanılacak tasarım spektrumlarının elde edilebilmesi, zemin tabakalarında oluşacak deplasman ve şekil değiştirmelerin belirlenebilmesi amacıyla serbest saha davranışı analizleri gerçekleştirilmektedir. Doğrusal, doğrusal olmayan veya eşdeğer doğrusal olarak gerçekleştirilebilen analizlerde zeminlerin genel olarak yatay tabakalandığı durumlarda tek boyutlu dinamik analizler yeterli olmaktadır. Serbest saha davranışları ile zemin cinsi, yerel zemin sınıfı ve anakaya derinliği gibi zemin koşullarının yanında maksimum yer ivmesi ve frekans içeriği gibi deprem özelliklerinin zeminlerin dinamik davranışına olan etkileri belirlenebilir. Bu çalışmada yerel zemin sınıflarının ve deprem özelliklerinin tek boyutlu dinamik davranış üzerindeki etkilerinin belirlenebilmesi amaçlanmıştır. Bu amaçla 16 adet anakaya mostrasında alınan kuvvetli yer hareketi ivme kaydı ile farklı yerel zemin sınıfına sahip 20 zemin modeli kullanılarak bir boyutlu saha tepki analizleri gerçekleştirilmiştir. 1 boyutlu dinamik analizler Deepsoil v6.1 yazılımında frekans tanım alanında eşdeğer doğrusal ve zaman tanım alanında doğrusal olmayan yöntemler kullanılarak gerçekleştirilmiştir. Zemin modelleri TBDY 2018 kapsamında anakaya özelliklerinin farklı olduğu ZB yerel zemin sınıfına dahil olan 2 adet kaya, ZC yerel zemin sınıfına dahil olan birer adet kum ve kil, ZD ve ZE yerel zemin sınıflarına dahil olan ve anakaya derinliğinin etkilerini görebilmek için oluşturulan dörder adet kum ve killerden oluşmaktadır. Depremlere ait maksimum yer ivmesi değerlerinin zeminlerin dinamik davranışı üzerindeki etkilerini görebilmek amacıyla analizlerde kullanılan 16 deprem ivme kayıtları her biri dörder tane olmak üzere 0.1 g, 0.2 g, 0.3 g ve 0.4 g maksimum yer ivmesi değerlerine ölçeklendirilmiştir. Ölçeklendirilen bu kayıtlar 20 farklı zemin modeline TBDY 2018'e göre yerel zemin sınıfı ZB ve ZA olan anakaya tabakasından etkitilmiştir. Eşdeğer doğrusal ve doğrusal olmayan analizler olarak gerçekleştirilen bir boyutlu dinamik analizler sonucunda her zemin modeli için yüzey tepki spektrumları, hakim periyot değerleri, en büyük spektral zemin büyütmesi değerleri, zemin tabakalarında meydana gelen deplasman ve şekil değiştirmelerin derinlikle değişimi elde edilmiştir. Ayrıca TBDY 2018 kullanılarak maksimum spektral ivme değerleri 0.1 g, 0.2 g, 0.3 g ve 0.4 g olan 4 ayrı noktadan alınan tasarım spektrumları ile yapılan analizler sonucunda elde edilen yüzey tepki spektrumları karşılaştırılmıştır. Gerçekleştirilen analizlerin sonuçları incelendiğinde maksimum yer ivmesi ile zemin tabakalarında oluşan şekil değiştirme ve yüzeyde meydana gelen deplasmanlar arasında doğrusal bir ilişki olduğu görülmektedir. Maksimum yer ivmesi değerleri ile spektral zemin büyütmeleri arasında belirgin bir ilişki kurulamazken zeminlerin rijitlikleri azaldıkça büyütmelerin ve periyotların yükseldiği saptanmıştır. Ayrıca aynı yerel zemin sınıfına sahip kum zeminlerde deplasman ve şekil değiştirmenin, kil zeminlerde ise spektral zemin büyütmesinin daha yüksek olduğu belirlenmiştir.

Earthquakes could affect the properties of the soil layers by causing cyclic loading, create displacements on the surface and damages to the superstructures. Therefore, the dynamic behavior of soils during strong ground motion is important in earthquake-safe construction design. Dynamic behavior of the soil layers is defined by the stress-strain properties depending on the deformation level that occurs under cyclic loads. Stress-strain behavior of soils and dynamic properties of soils like shear wave velocity (Vs), damping ratio (D), and shear modulus (G) are determined using various laboratory and field experiments. In addition to the effects of earthquakes on the soil layers, as the waves move from the bedrock to the ground surface, they interact with the soil layers. Stress increases due to earthquake motion can cause loss of strength, collapses, high deformations and liquefaction in the soil layers. At the same time, the properties of the soils affect the waves traveling to the ground, leading to an increase in their amplitude. Dynamic ground response analyzes are carried out in order to obtain the design spectra which are used in superstructure designs and to determine the displacements and deformations that will occur in the soil layers. Ground response analysis can be performed as a linear, nonlinear, or equivalent linear analysis. In the linear analysis, the soil is defined as a material with Kelvin-Voigt solid properties and with a constant shear damping ratio and shear modulus. However, soils are not materials that exhibit linear stress-strain behavior. For this reason, it is necessary to use nonlinear analyzes in the time domain, especially at high strain levels, or equivalent linear analyzes in the frequency domain where nonlinear parameters are input by iterations. The methods used in ground response analysis are carried out in one dimensional (1-D), two dimensional (2-D), and three dimensional (3-D). In areas where topographic effects and region geometry are prominent, it is recommended to select 2-D or 3-D analyzes, whereas in cases the soil layers are generally horizontal, one dimensional dynamic ground response analyzes are sufficient. Analyses in all dimensions have boundary conditions according to their extent. In one-dimensional analysis, it is assumed that the waves travel from the bedrock to ground surface in one-dimension and the soil layers settle horizontally unlimitedly. Through ground response analysis, besides the effects of earthquake properties such as peak ground acceleration (PGA) and frequency content, effects of local soil conditions as soil type, site class and bedrock depth on the dynamic behavior of soils can be determined. In the Turkish Building Earthquake Resistant Design Regulation (TBDY) published in 2018, local site classes are determined by taking into account the regional characteristics of the soils as shear wave velocity, standard penetration test values and undrained shear strength. In line with TBDY, according to the shear wave velocity local site classes are ZA for higher than 1500 m/s, ZB for between 760 m/s-1500 m/s, ZC for between 360 m/s-760 m/s, ZD for between 180 m/s-360 m/s and ZE lower than 180 m/s. In addition, dynamic ground response analyzes have been made compulsory in areas included in the ZF local site class. In TBDY, nonlinear analysis is recommended where the strain level exceeds 1%. In this study, it was aimed to determine the effects of local soil conditions and earthquake properties on one-dimensional dynamic soil behavior. For this purpose, one-dimensional ground response analyzes were carried out using strong ground motion acceleration time history records taken from 16 bedrock outcrops and soil models with different local site classes. 20 models were used in the study. Firstly, in order to see the effect of local site class difference on the analysis results, 8 models were determined as their shear wave velocity values for 30 m thick soil layers near the surface to be included in the local site class ZB, ZC, ZD and ZE according to TBDY 2018. These models are 2 layers rocks which are included in the ZB local site class and had different bedrock properties, 3 sandy soils and 3 clayey soils which are included in the ZC, ZD and ZE local site classes. In order to see the effect of bedrock depth on the analysis results, 12 new models have been created in profiles with ZD and ZE local soil class with a bedrock depth of 30 m, 35 m, 40 m and 50 m and the number of models has been increased to 20. Thus, 2 rocks, 9 sandy soils and 9 clayey soils models were obtained. In order to see the effects of peak ground acceleration values of the earthquake motion on the dynamic behavior of the soils, the 16 earthquake acceleration records used in the analyzes were scaled to 0.1 g, 0.2 g, 0.3 g and 0.4 g peak ground acceleration values, for each. Chichi, Coyote, Nahanni and Norhridge-2 earthquake acceleration histories were included D1 earthquake group by scaling 0.1 g PGA value, Imperial Valley, Kocaeli, Loma Prieta and Whittier Narrows acceleration records were included D2 earthquake group by scaling 0.2 g PGA value, Düzce, Northridge, Parkfield and San Fernando acceleration records were scaled to 0.3 g PGA value and included in the D3 earthquake group, Loma Prieta-2, Mammothlake, Manjil and Tottori acceleration records were included in the D4 earthquake group by scaling to 0.4 g PGA value. The magnitude of the randomly selected earthquake records varies between 5.7 Mw and 7.6 Mw. Besides peak ground accelerations, Düzce earthquake acceleration record stands out with 1.17 m/s in peak ground velocity values. If the Arias Intensity values were used in the interpretation of the earthquake effects were examined, it is seen that the Arias Intensity of Imperial Valley and Düzce acceleration records in the group D3 were high as well as the earthquakes records included group D4. In the analysis, the bedrock was selected as elastic bedrock to actualize the conditions of ZB and ZA local site class according to TBDY. In the models used in the analyzes, the unit weight for bedrock was selected as γ=22.0 kN/m3. The shear wave velocity values of the bedrock in the models with rock were 1000 m/s and 1500 m/s, in other models it is 760 m/s. Shear wave velocity values for 30 m thick layers were chosen as 810 m/s for models with local site class ZB, 560 m/s for models with local site class ZC, 270 m/s for models with local site class ZD, and 150 m/s for models with local site class ZE. Shear wave velocities were chosen as the mean of the boundaries in the models where the local site classes were ZC and ZD, near the lower limit in the models with ZB, near the upper limit in the models with ZE. Shear wave velocity, unit weight and local site classes were kept the same on sandy and clayey soils, so the effect of soil type on the results was investigated. The selected materials were classified as dense, medium and loose sand, and stiff, medium plasticity and soft clay with respect to Turkish Building Earthquake Regulation. Relative density values for sand were 5%, 15% and 40%, respectively, and plasticity index value for clays was 5%, 15% and 40%, respectively. One-dimensional dynamic ground response analyzes were carried out as equivalent linear analysis in frequency domain and nonlinear analysis in time domain using 20 models and 16 scaled earthquake acceleration records. Within the scope of the study, approximately 800 one-dimensional analyses were performed in Deepsoil V6.1, a C ++ based seismic response analysis program. Deepsoil is a one dimensional analysis program with a self-defined degredation curves that allows linear, nonlinear and equivalent linear analysis. While performing the analyzes, soil profiles were divided into 5 m thick sublayers in the Deepsoil program and parameters were entered. As a result of the analyzes, surface response spectra, dominant period values, maximum spectral soil amplification values, displacements and deformations occurring in the soil layers were obtained for each model. The results obtained were evaluated according to the analysis made by equivalent linear analysis or nonlinear analysis, according to the differences of the local site classes, according to the soil layers being composed of clay or sand, according to the thickness of the transition layers under the 30 m thick layers and the change of the peak ground acceleration. Then design spectra were drawn according to the TBDY for ZB, ZC, ZD and ZE types of soil and for 4 different points with peak ground acceleration values 0.1 g, 0.2 g, 0.3 g and 0.4 g and earthquake level for DD-2 (with probability of exceeding the spectral accelerations over 50 years is 10% and the return period is 475 years). These design spectra were compared with the surface response spectra obtained from the analyzes of 16 bedrock outcrops. As a result of the analyzes performed with models were created to determine the effect of bedrock depth and having ZD and ZE local site classes, it is seen that the bedrock depth has no significant effect on spectral soil amplification. The displacements obtained on the surface and strains occurred in the soil layers are examined. It was seen that these values increased with bedrock depth. In the analysis carried out to determine the effect of bedrock properties, the maximum soil amplification was 1.9 when the bedrock was in the local site class ZA, and 1.7 in the case of ZB. Soil amplification increased as the impedance ratio between bedrock and soil layers increased. The maximum average displacement values obtained on the surface for the earthquake groups D1, D2, D3, and D4 were 2.7 cm, 2.9 cm, 1.8 cm and 4.3 cm respectively as a result of the equivalent linear analyses and, 2.6 cm, 2.9 cm, 1.9 cm and 4.4 cm as a result of nonlinear analysis. Maximum average strains occurred in soil layers for the earthquake groups were 0.14%, 0.18%, 0.13%, 0.29% for equivalent analyses and 0.20%, 0.24%, 0.16%, 0.38% for nonlinear analyses. It can be said that as the PGA value increased the deformation occurring in the soils increased when strain and displacement values were evaluated together. There was not a certain relationship between the change of peak ground acceleration and soil amplification but in general, higher soil amplifications were observed in earthquakes with lower PGA values. Maximum spectral soil amplification values were obtained as 3.3 in clays where the local site class was ZE. This value was 2.2 for sands at the same site class. While maximum spectral soil amplification was in 1.0 s period in clays, it was in 1.7 s period in sands. From this, it can be concluded that soil amplification is higher in clays, but sands change the frequency content of the earthquake movement more. In cases where local site classes were the same, the maximum displacement on the surface was on average 4.1 cm in clays and 4.3 cm in sands. Besides, the maximum strain values are 0.46% on sands, and 0.20% on clays. In general, if the local site class, shear wave velocities and unit weights are the same for sands and clays, sands are exposed to more deformation, while the clays enlarge the earthquake amplitudes more. In adition to this, when the surface response spectra obtained as a result of the equivalent linear and nonlinear analyzes and the design spectrum determined within the scope of TBDY are compared, there is not major difference about spectral accelearations and dominant periods except for those obtained for sandy soils where the local soil site is ZE. In models created with sandy soils where the local site class is ZE, the values in the acceleration spectra were lower in the spectral acceleration values in the design spectra determined within the scope of TBDY.