The properties of natural soil are inherently variable and influence design decisions in geotechnical engineering. Apart from the inherent variability of the soil, the variability may arise due to measurement of soil properties in the field or laboratory tests and model errors. These wide ranges of variability in soil are expressed in terms of mean, variance and autocorrelation function using probability/reliability based models. The most common term used in reliability based design is the reliability index, which is a probabilistic measure of assurance of performance of structure. The main objective of the reliability based design is to quantify probability of failure/reliability of a geotechnical system considering variability in the design parameters and associated safety. In foundation design, reliability based design is useful compared to deterministic factor of safety approach. Several design codes of practice recommend the use of limit state design concept based on probabilistic models, and suggest that, development of reliability based design methodologies for practical use are of immense value. The objective of the present study is to propose reliability based design methodologies for pile foundations under static and seismic loads. The work presented in this dissertation is subdivided into two parts, namely design of pile foundations under static vertical and lateral loading; and design of piles under seismic loading, embedded in non-liquefiable and liquefiable soil. The significance of consideration of variability in soil parameters in the design of pile foundation is highlighted. A brief review of literature is presented in Chapter 2 on current pile design methods under vertical, lateral and seismic loads. It also identifies the scope of the work. Chapter 3 discusses the methods of analysis which are subsequently used for the present study. Chapter 4 presents the reliability based design methodology for vertically and laterally loaded piles based on cone penetration test data for cohesive soil. CPT data from Konaseema area in India is used for analysis. Ultimate limit sate and serviceability limit state are considered for reliability based design using CPT data and load displacement curves. Chapter 5 presents the load resistance factor design (LRFD) of vertically and laterally loaded piles based on load test data. Reliability based code calibrated partial factors are determined considering bias in failure criteria, model bias and variability in load and resistance. Chapter 6 illustrates a comprehensive study on the effect of soil spatial variability on response of vertically and laterally loaded pile foundations in undrained clay. Two-dimensional finite difference program, FLAC2D (Itasca 2005) is used to model the soil and pile. The response of pile foundations due to the effect of variance and spatial correlation of undrained shear strength is studied using Monte Carlo simulation. The influence of spatial variability on the propagation and formation of failure near the pile foundation is also examined. Chapter 7 describes reliability based design methodology of piles in non-liquefiable soil. The seismic load on pile foundation is determined from code specified elastic design response spectrum using pseudo-static approach. Variability in seismic load and soil undrained shear strength are incorporated. The effects of soil relative densities, pile diameters, earthquake predominant frequencies and peak acceleration values on the two plausible failure mechanisms; bending and buckling are examined in Chapter 8. The two-dimensional finite difference analysis is used for dynamic analysis. A probabilistic approach is proposed to identify governing failure modes of piles in liquefiable soil in Chapter 9. The variability in the soil parameters namely SPT-N value, friction angle, shear modulus, bulk modulus, permeability and shear strain at 50% of modulus ratio is considered. Monte Carlo simulation is used to determine the probability of failure. A well documented case of the failed pile of Showa Bridge in 1964 Niigata earthquake is considered as case example. Based on the studies reported in this dissertation, it can be concluded that the reliability based design of pile foundations considering variability and spatial correlation of soil enables a rational choice of design loads. The variability in the seismic design load and soil shear strength can quantify the risk involved for pile design in a rational basis. The identification of depth of liquefiable soil layer is found to be most important to identify failure mechanisms of piles in liquefiable soil. Considerations of soil type, earthquake intensity, predominant frequency of earthquake, pile material, variability of soil are also significant.

The official recommendations in geotechnical engineering are according to Swedish practice to install inclined driven piles in order to resist external horizontal loads, which usually consist of windloads. These loads gets counteracted by utilizing the pile's axial load-bearing capacity, however when designing a foundation the use of inclined piles has proven to be problematic from a number of different perspectives. This has made both engineers as well as contractors long for a solution where the piles instead are allowed to be installed vertically, which could be made possible by utilizing the lateral bearing capacity of the pile which occurs in connection with lateral resistance during pile-soil interaction. The present day knowledge about such an engineering procedure in Sweden has proven to be limited and consists mainly of one governing document, namely report 101 published by the Commission on Pile Research. The aim of our study is to test and evaluate the method in report 101 based on a number of different load cases related to lateral load effects during pile-soil interaction. An analytical method reproduced from the Commission on Pile Research's report has been implemented with help of Mathcad, a computer software for reuse of mathematical calculations. Decisive parameters that distinguish the different load cases have included external load impact in pile head, mechanical strength properties of friction or cohesive soil, along with cross-sectional dimensions and reinforcement content of piles etc. The results differed significantly depending on the loadcase, a majority of the given answers were also perceived as unreliable and in a number of load cases the method also failed to obtain an analytical solution. Our conclusion is that it is doubtful whether the elastoplastic method presented in report 101 in the end in practice benefits the engineer tasked with designing, this stands in stark contrast to an increasing demand for easily accessible knowledge within its field. / Pålar i samband med grundläggning har sedan en tid tillbaka enligt svensk praxis installerats snedslagna, detta för att med hjälp av pålens axiella bärförmåga motstå yttre horisontell lastpåverkan till följd av exempelvis vindlaster. Att snedställa pålar har visat sig vara problematiskt ur ett flertal olika perspektiv, en lösning som både konstruktörer och entreprenörer efterlängtat är att pålarna istället tillåts installeras rakställda. Detta skulle vara möjligt genom att nyttja pålens transversella bärförmåga som uppstår i samband med sidomotstånd vid interaktion påle-jord. Kunskap kring ett sådant dimensioneringsförfarande i Sverige idag har visat sig vara begränsad och till stor del bestå av ett styrande dokument, nämligen rapport 101 utgiven av Pålkommissionen. I denna studie har en analytisk handberäkningsmetod som återges ur Pålkommissionens rapport implementerats i beräkningsprogrammet Mathcad, detta i syfte att testa och utvärdera metoden utifrån ett flertal olika belastningsfall relaterade till transversell lastpåverkan vid interaktion påle-jord. Avgörande parametrar som skiljt de olika belastningsfallen åt har bl.a. utgjorts av hållfasthetsegenskaper hos friktion- eller kohesionsjord, tvärsnittsdimensioner och armeringsinnehåll hos påle, yttre lastpåverkan i påltopp. Resultaten skiljde sig åt markant beroende på typ av belastningsfall där en majoritet av svaren upplevdes som icke tillförlitliga och till ett flertal olika fall, utifrån vårt försök till implementering, klarade metoden heller inte av att erhålla någon analytisk lösning. Vi finner det därför svårt att tro att den elastoplastiska metoden som återges i rapport 101 i slutändan kommer till någon direkt praktisk nytta, vilket står i tvär kontrast mot ett uppenbart behov där lättillgänglig kunskap inom området efterfrågas.

A series of static lateral load tests were conducted on a group of fifteen piles arranged in a 3x5 pattern. The piles were placed at a center-to-center spacing of 3.92 pile diameters. A single isolated pile was also tested for comparison to the group response. The subsurface profile consisted of cohesive layers of soft to medium consistency underlain by interbedded layers of sands and fine-grained soils. The piles were instrumented to measure pile-head deflection, rotation, and load, as well as strain versus pile depth.