This thesis consists of the development, trial an application of a computational tool that by the use of the Genetic Algorithm optimization method provides the user with an alternative approach from the traditional trial and error when faced with the challenge of designing end bearing pile groups.The work presents the three main concepts needed for understanding the computational tool as well as the structure of it. The scripting capabilities of a FEM software are exploited in order to automatically generate several design alternatives that are then compared in quality in an iterative way. The developed algorithm succeeds validation against pile group problems with known solutions and is then applied to the analysis of three different real structures with different levels of complexity.The algorithm is suitable for optimizing problems with loading in three directions and several load cases, within a time range consistent with an overnight run, when the density of the pile group is not high. Another potential application is to automatize retaining wall resting on piles type problems within a short amount of computational time. / Denna avhandling består av utveckling och prövning av ett beräkningsverktyg som med hjälp av den genetiska algoritmoptimeringsmetoden ger användaren ett alternativt tillvägagångssätt från den traditionella prövningen när den möter utmaningen att utforma bärande höggrupper.Arbetet presenterar de tre huvudkoncepten som behövs för att förstå beräkningsverktyget och dess struktur. Skriptfunktionerna för en FEM-programvara utnyttjas för att automatiskt generera flera designalternativ som sedan jämförs i kvalitet på ett iterativt sätt. Den utvecklade algoritmen lyckas validera mot höggruppsproblem med kända lösningar och tillämpas sedan för analys av tre olika byggda strukturer med olika nivåer av komplexitet.Algoritmen är lämplig för att optimera problem med belastning i tre riktningar och flera lastfall, inom ett tidsintervall som överensstämmer med en löpning över natten, när täthetsgruppens densitet inte är hög. En annan mycket intressant applikation är att automatisera kvarvarande väggproblem inom en kort tid av beräkningstid.

The soil-structure interaction models associated with laterally loaded deep foundations have typically been based on load tests involving relatively small diameter foundations. The lateral soil resistance for larger diameter foundations has been assumed to increase linearly with diameter; however, few, if any load tests have been performed to confirm this relationship. To better understand the lateral resistance of large diameter deep foundations in sand, a series of full scale, cyclic, lateral load tests were performed on two 1.2 m diameter drilled shafts and a 0.324 m diameter steel pipe pile in sand. Although the tests involve two different foundation types, the upper 2.4 m of the profile, which provides the majority of the lateral resistance, consists of sand compacted around both foundation types. Therefore, these test results make it possible to evaluate the effect of foundation diameter on lateral soil resistance. The drilled shafts were first loaded in one direction by reacting against a fifteen-pile group. Subsequently a load test was performed in the opposite direction by reacting against a 9-pile group. The soil profile below the 2.4 m-thick layer of compacted sand consisted of interbedded layers of sand and fine-grained soil. For the drilled shaft load tests, pile head deflection and applied load were measured by string potentiometers and load cells, respectively. Tilt was also measured as a function of depth with an inclinometer which was then used to calculate deflection and bending moment as a function of depth. For the pipe pile, deflection and applied load were also measured; however, bending moment was computed based on strain gauges readings along the length of the pile. The lateral response of the drilled shafts and pipe pile were modeled using the computer programs LPILE (Reese et al., 2000), SWM6.0 (Ashour et al., 2002), and FB-MultiPier Version 4.06 (Hoit et al., 2000). Comparisons were made between the measured and computed load-deflection curves as well as bending moment versus depth curves. Soil parameters in the computer programs were iteratively adjusted until a good match between measured and computed response of the 0.324 m pipe pile was obtained. This refined soil profile was then used to model the drilled shaft response. User-defined p-multipliers were selected to match the measured results with the calculated results. On average very good agreement was obtained between measured and computed response without resorting to p-multipliers greater than 1.0. These results suggest that a linear increase in lateral resistance with foundation diameter is appropriate. LPILE typically produced the best agreement with measured response although the other programs usually gave reasonable results as well. Cyclic loading generally reduced the lateral resistance of the drilled shafts and pile foundation by about 20%.

Two different concepts and design procedures namely settlement reducing piles and piled raft foundations have been studied independently in this thesis. A laboratory study is conducted on model rafts with differing number of model settlement reducing piles. Pile length, pile diameter, type of soil and size of raft are kept constant and settlements are measured under sustained loading. Remolded kaolin is consolidated under controlled stresses before tests are performed in model boxes. The tests are conducted under two sustained loadings of 75 kPa and 40 kPa. 0(raft), 16 and 49 number of piles are used. During the tests, all of the skin friction is mobilized. Several tests are conducted for each combination to see the variability. It is concluded that increasing the pile number beyond an optimum value is inefficient as far as the amount of settlement is considered. Also an analytical procedure has been followed to calculate settlements with increasing number of piles. In the second part of this thesis, finite element analyse have been performed on a piled raft foundation model, using Plaxis 3D Foundation Engineering software. This analyse are supported with analytical methods. The piled raft model is loaded with 450 kPa raft pressure. The studies are conducted in two sets in which different pile lengths are used / 25 m and 30 m respectively. The numbers of piles are increased from 63 to 143. All other parameters are kept constant. The results showed that again an optimum number of piles will be sufficient to reduce the settlement to the acceptable level. The analytical methods indicate a similar behavior. The comparison and results are presented in the study.

To investigate the lateral load distribution of each pile in a pile group, the bending moment distribution along the pile and the lateral group displacements with respect to pile location in the group, pile spacing, pile diameter and soil stiffness three dimensional finite element analysis were performed on 4x4 pile groups in clay. Different Elatic Modulus values, pile spacings, pile diameters and lateral load levels used in this study. In the analysis PLAXIS 3D Foundation geotechnical finite element package was used. It is found that, lateral load distribution among the piles was mainly a function of row location in the group independent from pile spacing. For a given load the leading row piles carried the greatest load. However, the trailing row piles carried almost the same loads. For a given load, bending moment values of the leading row piles were greater than the trailing row piles. On the other hand, as the spacing increased group displacements and individual pile loads decreased under the same applied load. However, this behavior was seen more clearly in the first and the second row piles. For the third and the fourth row piles, pile spacing became a less significant factor affecting the load distribution. It is also found that, pile diameter and soil stiffness are not significant factors on lateral load distribution as row location and pile spacing.

Much research has been done to study the effects of spacing in laterally loaded pile groups and how the pile spacing affects lateral resistance of the piles. In this test a 9-pile group of steel piles was installed in a 3x3 configuration in sand. The piles were spaced from center-to-center at a distance of 5.65 pile diameters in the direction of the load. The pile group was laterally loaded and instrumented to collect deflection, load, and strain data. A separate single pile was similarly instrumented and tested for comparison to the pile group. The soil profile consisted of a top layer of sand about 2.5 meters deep underlain by alternating layers of fine grained soil and sand. Analysis was done in order to determine p-multipliers for the rows in the pile group. The pile group consistently resisted lower average loads than those of the single pile at the same peak deflection. Row 1 resisted equal loads to those resisted by the single pile but Row 2 and Row 3 resisted smaller loads successively. The maximum bending moments were greatest in Row 1 and decreased successively in Row 2 and Row 3, however they occurred at the same depth for the same peak deflection. Maximum bending moments for the single pile were similar to those for Row 1 in the pile group at each deflection. Group effects were more significant at greater loads with larger deflections. Higher deflection caused increased shear zone interaction and a decrease in lateral resistance. The test on the single pile was modeled using the computer program LPILE Plus version 4.0 (Reese et al., 2000). The soil profile was calibrated by alternating the input soil parameters until the curves matched between the measured and the calculated results. The new soil profile was then used to model the test on the pile group using GROUP version 4.0 (Reese and Wang, 1996). A match was made between the measured test results and those calculated using GROUP by defining p-multipliers for each of the rows. For Cycle 1 the p-multipliers were found to be 1.0, .55, and .5 for Row 1 through Row 3 respectively. For Cycle 10 the p-multipliers were found to be the same as those for Cycle 1.

Submitted by Luciana Ferreira (lucgeral@gmail.com) on 2016-04-20T11:13:08Z No. of bitstreams: 2 Dissertação - Tallyta da Silva Curado - 2015.pdf: 5629918 bytes, checksum: 52dabea4c56416cc72708f0af09a3640 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2016-04-20T11:16:53Z (GMT) No. of bitstreams: 2 Dissertação - Tallyta da Silva Curado - 2015.pdf: 5629918 bytes, checksum: 52dabea4c56416cc72708f0af09a3640 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Made available in DSpace on 2016-04-20T11:16:53Z (GMT). No. of bitstreams: 2 Dissertação - Tallyta da Silva Curado - 2015.pdf: 5629918 bytes, checksum: 52dabea4c56416cc72708f0af09a3640 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2015-08-27 / Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico - CNPq / With the expansion of urban areas, there is a tendency to vertical structures, with these being increasingly slim and constructed into smaller areas. In these cases, for it is often necessary to measure deep foundations, as piled foundations in order to resist high internal forces and with little available space, resulting in caps with many piles and next to each other, or even forming piled rafts. The influence of this high amount and proximity of piles cannot be neglected. To consider the effect of a pile in another, several theories have been formulated to better represent the interaction between piles, but there is not yet an established theory which results in factors that best represent the reality. In this way, this work aims to study existing methods in predicting the interaction factor between piles comparing results of analytical methods with numerical solutions. For this comparison was used the program of finite element method DIANA as a reference the values considered closer to the exact solution. The various theories are compared in accordance with the cases already reported in the literature, through the factors interaction between piles. Thus, it can be seen that the theories of Cao and Chen (2008) and Mylonakis Gazetas (1998) showed good agreement with the results of DIANA, but not in all cases of combination relative length (L/d) and relative stiffness (K), as Poulos and Davis (1980). And when one considers the possibility of having end-bearing piles, both Poulos and Davis (1980) and El Sharnouby and Novak (1990) overestimate, largely, the values from the finite element solution. However, this last is more precise in the condition of layered soil than the first theory. Furthermore, analyzes were performed by modeling with varying amounts of piles, in order to evaluate the influence of the presence of intermediate piles in the interaction between the piles. Thus, it was noted that there was increasing variation of factors interaction between piles with increasing amount of piles, such as when the piles are arranged concentrically and not more in line manner. Finally, settlements of pile groups are analyzed, in order to verify the difference between the settlements obtained via modeling by DIANA and through prediction method of settlement of pile groups. The latter analysis showed that small differences between the interaction factors between two piles could change the forecast of settlements by theoretical methods, causing representing differences. / Com a expansão dos centros urbanos, verifica-se a tendência de verticalização das estruturas, com estas sendo cada vez mais esbeltas e construídas em áreas menores. Com isso, por muitas vezes é necessário dimensionar fundações profundas, como as estaqueadas, de modo a resistir a altos esforços solicitantes e com pequena disponibilidade de espaço, implicando em blocos com muitas estacas, e próximos uns aos outros, ou mesmo formando radiers estaqueados. A influência dessa elevada quantidade e proximidade de estacas não pode ser desprezada. Para considerar o efeito de uma estaca em outra, várias teorias vêm sendo formuladas para representar melhor a interação entre estacas, mas ainda não há uma teoria estabelecida como a que resulta em fatores que representam melhor a realidade. Neste caminho, esse trabalho tem por objetivo estudar as metodologias existentes na previsão do fator de interação entre estacas comparando resultados de métodos analíticos com soluções numéricas. Para essa comparação foi utilizado o programa de Método dos Elementos Finitos DIANA como uma referência dos valores considerados mais próximos da solução exata. As diversas teorias são comparadas, de acordo, com os casos já apresentados na literatura, por meio dos fatores de interação entre estacas. Com isso, pode-se verificar que as teorias de Cao e Chen (2008) e Mylonakis e Gazetas (1998) tiveram boa concordância com os resultados do DIANA, mas não em todos os casos de combinação de comprimento relativo (L/d) e rigidez relativa (K), assim como Poulos e Davis (1980). E quando se considera a possibilidade de se ter estacas de ponta, tanto Poulos e Davis (1980) quanto El Sharnouby e Novak (1990) superestimam, em grande parte, os valores provenientes da solução de elementos finitos. Entretanto, esta última é mais precisa na condição de solo estratificado do que a primeira teoria. Ainda, foram realizadas análises por meio de modelagens com variadas quantidades de estacas, de modo a avaliar a influência da presença de estacas intermediárias na interação entre as estacas. Com isso, percebeu-se que há variação crescente dos fatores de interação entre estacas com o aumento da quantidade destas, assim como quando as estacas são dispostas de maneira mais concêntrica e não em linha. Por fim, recalques de grupos de estacas são analisados, visando verificar a diferença entre os recalques obtidos via modelagem pelo DIANA e por meio de método de previsão de recalque de grupos de estacas. Esta última análise demonstrou que diferenças pequenas entre fatores de interação entre duas estacas, podem alterar a previsão de recalques por métodos teóricos, provocando diferenças representativas.

This thesis has two major goals; first to develop scalable scripts for steady-state analysis, then to perform a case study on the dynamic properties of a vertical pile. The scripts are based on the numerical library PETSc for parallel linear algebra. This opens up the opportunity to use the scripts to solve large-scale models on supercomputers. The performance of the scripts are verified against problems with analytical solutions and the commercial software ABAQUS. The case study compares the numerical results with those obtained from an approximate solution.   The results from this thesis are verified scripts that can find a steady-state solution for linear-elastic isotropic solids on supercomputers. The case study has shown differences between numerical and semi-analytical solutions for a vertical pile. The dynamic stiffness show differences within reasonable limits but the equivalent viscous damping show larger differences. This is believed to come from the material damping in the soil that has been excluded from the approximate solution.   These two results make it possible for further case studies on typical three-dimensional problems, that result in large-scale models, such as the dynamic properties of a slanted pile or pile-groups. The scripts can easily be expanded and used for other interesting research projects and this is the major outcome of from this thesis.

[pt] Este trabalho tem como objetivo apresentar um método para previsão de recalque de grupos de estacas, solicitadas axialmente em maciços de solo estratificados, através do emprego de técnicas numéricas. O método de análise utilizado é baseado em uma versão simplificada do método dos elementos de contorno, sugerida por Poulos e Davis (1968) para o caso de solo homogêneo, isotrópico e linear elástico, desenvolvida a partir das soluções de Mindlin (1936). Neste caso, a modificação proposta constou da substituição de meio estratificado por meios homogêneos equivalentes, através da utilização de procedimentos sugeridos por Poulos (1979) e Yamashita et al (1987). Os valores de recalque obtidos com os programas computacionais desenvolvidos nesta pesquisa foram comparados com resultados correspondentes obtidos pelo métodos dos elementos finitos ou publicados na literatura por diferentes autores e através de diversas metodologias. Estas comparações permitem concluir que a utilização da proposição aproximada sugerida neste trabalho fornece, sob certas condições, valores bastante satisfatórios do recalque de grupos de estacas em maciços de solo estratificados. As vantagens de utilização deste procedimento estão diretamente associadas a uma significativa simplificação na preparação dos dados para as necessárias descrições geométricas e geotécnicas do problema bem como a redução do tempo de processamento. Fatores de interação para grupos de estacas foram produzidos para alguns casos específicos de grupos em maciços de solo formados por camadas duplas. / [en] This thesis presents a numerical analysis of pile groups deflections, under axial load in layered soils. The method uses a simplified version of the boundary element technique, as proposed by Poulos and Davis (1968) for a homogeneous, isotropic and linear elastic half-space, based on Mindlin s equation. In this case, the major step consisted in replacing the multi-layered system for an equivalent homogeneous half-space using the well known procedures suggested by Poulos (1979)and Yamashita et al (1987). Results of the deflection of pile groups obtained with the proposed method were compared to results obtained using finite element analysis and to results published in the technical literature obtained by several authors. It has been shown that the proposed method provides results which compare favorably with existing solution and basically presents the advantage of the simplification in handling the soil and geometrical datas necessary for the problem description. Interaction factors for groups of pile were produced for some specific cases consisting of a double layer system. / [es] Este trabajo tiene como objetivo presentar un método para previsión de recalque de grupos de estacas, solicitadas axialmente en macizos de suelo extratificados, a través del empleo de técnicas numéricas. EL método de análisis utilizado tiene como base en una versión simplificada del método de los elementos de contorno, sugerida por Poulos y Davis (1968) para el caso de suelo homogéneo, isotrópico y lineal elástico, desarrollada a partir de las soluciones de Mindlin (1936). En este caso, la modificación propuesta incluye la substituición de medio extratificado por medios homogéneos equivalentes, a través de la utilización de procedimentos sugeridos por Poulos (1979) y Yamashita et al (1987). Los valores de recalque obtenidos con los programas computacionales desarrollados en esta investigación fueron comparados con resultados correspondentes obtenidos por el métodos de los elementos finitos o publicados en la literatura por diferentes autores y a través de diversas metodologías. Estas comparaciones permiten concluir que la utilización de la proposición aproximada sugerida en este trabajo brinda, bajo certas condiciones, valores bastante satisfactorios del recalque de grupos de estacas en macizos de suelo extratificados. Las ventajas de utilización de este procedimiento están directamente asociadas la una significativa simplificación en la preparación de los datos para las necesarias descripciones geométricas y geotécnicas del problema así como la reducción del tiempo de procesamiento. Fueron produzidos factores de interación para grupos de estacas para algunos casos específicos de grupos en macizos de suelo formados por estratos duplas.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18558号 / 工博第3919号 / 新制||工||1602(附属図書館) / 31458 / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 平石 哲也, 教授 間瀬 肇, 准教授 馬場 康之 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

[pt] O comportamento de grupos de estacas sob carregamento vertical estático, cêntrico, com bloco rígido ou flexível, em maciços estratificados, é investigado neste trabalho. O método de análise é baseado em versão simplificada do Método dos Elementos de Contorno, sugerida por POULOS (1968) para meios elásticos homogêneos. Estimativas dos valores de recalque são obtidas por meio de um programa computacional, implementado nesta pesquisa, avaliando-se a influência da estratificação do maciço de solo nos efeitos de grupo e nos valores dos recalques do topo das estacas componentes. Grupos com diferentes configurações, usuais na prática da Engenharia de Fundações, foram considerados em perfis estratigráficos básicos, homogêneo e formado por duas camadas distintas. Os resultados numéricos obtidos para o caso de bloco flexível, indicam que a estratigrafia do maciço, a rigidez relativa das estacas e seus comprimentos relativos não influenciam significativamente os valores dos fatores de interação para cálculo dos recalques do topo. Assim, um método aproximado simples, direto e eficiente, é proposto para determinação dos recalques do topo em grupos com número qualquer de estacas, com base na superposição dos valores do fator de interação entre apenas duas estacas sob diferentes espaçamentos relativos. No caso de bloco rígido, os resultados obtidos indicaram que, sob o ponto de vista prático, os fatores de redução de grupo também não são significativamente influenciados pela estratigrafia, rigidez e comprimentos relativos das estacas. Dessa forma, os fatores de redução de grupo obtidos para o perfil de solo homogêneo podem ser aproximadamente utilizados para outros casos de estratigrafia. / [en] The behaviour of vertically loaded pile groups with a rigid or flexible cap, in layered soils, is investigated in this research. The analysis is based on a simplified version of the Boundary Element Method, as suggested by POULOS (1968), for the case of an homogeneous and linearly elastic half-space. The estimation of settlements is obtained using a computational program, developed in this work, which evaluates the influence of the soil layering on the group effects as well as on the settlement of individual piles in the same group. Groups with usual geometrical configuration are considered, resting on some basic soil profiles (single and double layered systems). Numerical results thus calculated for the case of flexible cap indicate that soil layering, pile-stiffness factors and length-to- diameter ratios do not significantly affect the interaction factors for the top of piles. Hence, a simple, direct and efficient method of evaluation is proposed for the estimation of settlements in groups with any number of piles based on the superposition of the interaction factors determined for the case of just two piles at different spacing. For groups with a rigid cap, under the pratical point of view, the results also indicated that, the respective reduction factors are not significantly influenced by the soil layering, pile-stiffness factors and length-to- diameter ratios. So, reduction factors obtained for homogeneous soil profile can be used for others cases of soil profiling. / [es] Este trabajo investiga el comportamiento de grupos de estacas bajo sobrepeso vertical estático, céntrico, con bloque rígido o flexible, en macizos estratificados. El método de análisis se basa en una versión simplificada del Método de los Elementos de Contorno, sugerida por POULOS (1968) para medios elásticos homogéneos. Estimativas de los valores de recalque se obtienen por medio de un programa, implementado en esta investigación, que evalúa la influencia de la estratificación del macizo de suelo en los efectos de grupo y en los valores de los recalques de la parte superior de las estacas. Grupos con diferentes configuraciones, usuales en la prática de Ingeniería de Fundaciones, fueron considerados en perfiles estratigráficos básicos, homogéneo y formado por dos capas distintas. Los resultados numéricos obtenidos para el caso de bloque flexible, indican que la estratigrafía del macizo, la rígidez relativa de las estacas y sus dimensiones relativas no influyen significativamente en los valores de los factores de interación para el cálculo de los recalques. De esta forma, se propone un método aproximado, simple, directo y eficiente para la determinación de los recalques en grupos con cualquier número de estacas, con base en la superposición de los valores del factor de interacción entre apenas dos estacas bajo diferentes espaciamientos relativos. En el caso de bloque rígido, los resultados obtenidos indicaron que, desde el punto de vista práctico, los factores de reducción de grupo también no recibenuna influencia significativa por parte de la estratigrafía, rígidez y dimensiones relativas de las estacas. De esa forma, los factores de reducción de grupo obtenidos para el perfil de suelo homogéneo pueden ser aproximadamente utilizados para otros casos de estratigrafía.