Development of finite element model for soil-structure interaction

The presence of soil in framed-structure analysis has shown great effects on the overall performance of soil-foundation-structure system. One of the important features of Soil-Structure Interaction (SSI) analysis is practical prediction of the ultimate bearing capacity of layered dried soil. In addition, mechanical properties of the soil interfacial behavior between soil and shallow infrastructures have been found influential to the structural behavior. Further to the necessity of an accurate interfacial modeling in the soil-foundation-framed structure system, the analyzed model should be computationally efficient to capture the structural performance by considering soil, footing and structure. Advances in the computer technology have led to the development of several commercial Finite Element (FE) softwares which are available to all researchers. However, none of them could serve as a comprehensive system to compute the ultimate bearing capacity of soil and its interactive effects on structural elements with an acceptable computational time and effort while considering interfacial behavior between soil and foundation. Therefore, the current research focuses on first, proposing Artificial Neural Network (ANN) as an advanced method capable of prediction of the ultimate bearing capacity of two-layer soil where water table is low enough so that there is no moisture effect on the failure surface. Secondly, an analytical finite element technique has been developed which is capable of modeling and analyzing soil-framed structure interaction equipped with new elements to accurately capture the different aspects of interaction. Hence, the architected ANN and the FE technique form an analysis system while their functionality is not dependent of each other and each one performs their computational tasks separately. In order to develop an efficient ANN system, the required algorithm is built through trial and error procedure and then through this process feed-forward network with back-propagation training function is implemented. The training data associated with ultimate bearing capacity of layered soil has been exploited from classical and numerical (FE) methods. For the classical method the well-known equation of Meyerhof for two-layered soil has been employed and a Two-Dimensional Finite Element Code (2DFEC) equipped with a multi-finite element library capable of soil non-linear analysis is utilized. Results from the FE and classical methods are compared for verification. In the present study, the SSI analytical model is idealized by the two-node beam element and the eight-node serendipity element. In order to consider interfacial behavior and tackle compatibility problem in modeling of SSI between soil and frame element, the new interface element is developed and the associated constitutive law and finite element model is derived. Then, the special finite element algorithm and computer program (2DSSI) is developed to perform two-dimensional nonlinear SSI analysis by considering interfacial behavior of soil and frame elements. Since different materials are used in the current SSI problem constitutive models are employed to capture both the linear and nonlinear interactive behavior of all the materials i.e. concrete, steel, soil and interface element. In order to evaluate the accuracy of the developed code (2DSSI), a series of examples have been used as benchmarks to verify the analyses. Furthermore, the accuracy of the program code is also verified through analysis of similar examples in literature and then results are compared with SAP2000 and those available in literature. Comparisons showed acceptable accuracy and reasonable prediction of the analysis. Results from ANN showed good agreement with the classical method. Therefore,ANN can be used as a powerful tool for prediction of ultimate bearing capacity of layered soil. On the other hand, application of the new interface element in SSI analysis has led to considerable changes and redistribution of induced forces and stresses in the frame elements, footing and soil. In general, application of interface element can well highlight the influence of soil behavior and its consequent effects on the superstructure. Presence of interface element gave rise to considerable redistribution and variation of structural forces. Depending on the mechanical properties of interface element (thickness and stiffness), geometry of superstructure and type of footing, the conducted analyses showed 3% to 38% degradation of axial forces of columns, 5% to 62% reduction of shear forces of columns, 1% to 87% and 6% to 96% redistribution of column and beam moments respectively compared to the models without the interface element. Moreover, Moment reversal and increase of values of forces in several structural elements were also remarkable in the considered cases that corresponded to the presence of interface element. Larger thickness of interface element added more flexibility to the combined footing and consequently larger settlement and remarkable variation of foundation moment were obtained. All the redistributions and alterations of structural forces contributed in changing the yielding mechanisms of the considered frames. Results revealed that inclusion of interaction between soil and footing along with additional interface element can cause redesign of structural members so the cost of construction may change. Nonlinear interactive analysis of an asymmetric frame in the current study showed 26% increase in the required volume of reinforced concrete for construction of the frame. The developed system, discussed above, includes an ANN algorithm to predict ultimate bearing capacity of two-layer soil and a finite element model (2DSSI) for analysis of interaction between soil and framed structures through utilization of the developed interface element along with all included aspects of SSI.

Preview PDF FK 2014 17R.pdf Download (1MB) | Preview

Actions (login required)

View Item