Determination of the effect of stratification on liquefacture of soils using numerical analysis

In the light of increasing scientific studies, many researches are being carried out on liquefaction, and one of the current topics is the examination of the upper layer effect in determining the liquefaction potential of soils. In this study, it is aimed to examine the stratification effect with the numerical analysis method in order to contribute to the literature on the preliminary evaluation chart suggested by Ishihara (1985) and the liquefaction severity index evaluations suggested by other researchers. In the studies carried out; It is aimed to evaluate the results obtained by using the numerical analysis program in determining the deformations that will occur in the structure above a non-liquefied layer in case there is a layer with liquefaction potential, in terms of relative storey drift, sliding, bending and settlement behaviors. First of all, in order to make the evaluation, a drilling log containing liquefied and non-liquefied layers in Sakarya/Adapazarı was selected and experimental parameters of the selected drilling log were obtained. A soil model was created, consisting of SM3 (without liquefaction potential) soil layer specified in the selected drilling log, on top, SM4 (with liquefaction potential) and CH4 clay layer underneath. First of all, one-dimensional analysis of the SK-2 column with a depth of 75 m within the selected log scope was carried out using the "1999 Marmara Earthquake record" in the literature, and it was observed that the SM3 layer did not liquefy, but the SM4 layer did. Before the one-dimensional (1D) analysis of the SK-2 drilling log in question, a point at -25 m elevation was defined and an acceleration record of the said point was obtained. When the acceleration record in the 25 m of the drilling log with a depth of 75 m was analyzed by modeling the first 25 m of the log in question and the results were compared, it was seen that there was a similarity. After the one-dimensional (1D) analysis, it was observed that when the same model was modeled as (2D), the acceleration values occurring on the ground surface increased by approximately 2 times. It has been determined that this situation may be caused by boundary conditions such as the seismic behavior occurring in the ground causing reflection in the ground and the model width not being kept short or long enough. Within the scope of investigating the upper layer effect, a building with a width of 10 m and a height of 15 m was modeled on the model created with the clay layer under the non-liquefiable layer and the liquefiable layer. Afterwards, the effects of the selected model width on the results were tried to be examined. In this context, models with different widths (600 m, 800 m and 1000 m) were made. Analyzes of the models were carried out. It was observed that in the model with a model width of 600 m, high acceleration values due to boundary conditions were seen in the middle of the model. It was observed that the pore water pressures occurring in the middle of the model decreased for the model with a model width of 800 m, and became more stable in the model with a model width of 1000 m. In order to reduce the effect of boundary conditions on the model, the model width was chosen as 1000 m. For numerical analysis, the UBC3D-PLM body model, which is available in the literature regarding modeling of behavior, was chosen. Relevant parameters were calibrated using the formulations in the user manual prepared by the relevant numerical analysis program. In the studies in the literature, it has been observed that the P_refvalue is accepted as 100 Kpa in order to be equal to atmospheric pressure. In this study, the P_refvalue, which is a model constant used within the scope of the UBC3D-PLM constitutive model, was calibrated for each layer in line with the equation suggested by Boulanger (2003b) for the K_σ correction coefficient relationship, and analyzes were carried out on the obtained values. In order to investigate the upper layer effect, numerical analyzes were carried out for a total of six models obtained by increasing the non-liquefied layer thickness in the first 10 m above the ground by 2 meters. Liquefaction severity index values used in the literature to determine liquefaction potential were determined by deterministic methods. Definitions of the obtained values were made. Analyzes were carried out in line with the determined soil profiles and necessary comparisons were tried to be made through the points determined on the model. Necessary comparisons were made on the points determined on the model in terms of relative storey drift, settlement, shifting and bending controls. First of all, when evaluated in line with the preliminary evaluation chart suggested by Ishihara (1985) in the literature, it was seen that it was compatible with the numerical analysis results obtained. It has been understood that if the liquefaction severity index scalings in the literature are evaluated together, they may lead to incorrect evaluation based on different concepts. It has been observed that the analysis results of the models modeled by Ishihara (1985) with the Abakta finite element software, which is based on the assumption that the volumetric deformation value reaches around 1% and is used only for preliminary evaluation, are compatible. An attempt was made to contribute to the conceptual evaluation of the results obtained in the liquefaction intensity scale evaluations obtained with deterministic methods. It should never be overlooked that the use of some liquefaction intensity scaling methods may not be suitable for the project to be implemented. It is seen that the selection of the damping ratio is greatly affected by the analysis results on a model basis, and also that the methods of determining the damping parameters in the literature are different from each other and in many studies, evaluations have been made to state that the studies on determining the damping values need to be improved. In the literature, it is recommended to define non-liquefied layers as HS-Small, but in this study, all layers were modeled with the UBC3D-PLM body model in order to simulate the effect of the non-liquefied layer on the non-liquefied layer. It is seen that the pore water pressure in the non-liquefied layer produces positive pressure and adds additional strength to the total effective stress. If this situation occurs in tight sand, it behaves in a way that causes volumetric expansion, so the calculated residual pore water pressures are considered to change sign as +/-. The acceleration values between the models were evaluated and compared using the PSA diagrams obtained by the program. It has been observed that there are large differences in terms of acceleration values reaching the surface between models without a liquefied layer and with a liquefied layer. It has been observed that this situation transfers acceleration values to a lesser extent due to the damping of the liquefied layer in the model at the acceleration value reaching the surface. Regarding the structure's settlement on the ground; It has been observed that as the thickness of the non-liquefied upper layer increases, the amount of settlement that occurs decreases. Determining the degree of damage in terms of relative storey drift; When damage assessments were made within the framework of the Turkish Building Earthquake Regulation (TBDY), a gradual assessment was made from damage to the infrastructure to severe damage as the thickness of the liquefied layer under the non-liquefied layer for the first 10 m increased. It is thought that if there is a liquefied layer in the first 10 m layer, its effect will have a high probability of reaching the surface.