Upper bound plastic limit analysis (PLA) solutions have been widely used to assess maximum capacity of laterally loaded piles and caissons. However, for the specific case of short piles and caissons with aspect ratios generally ranging from one to three, the current solutions tend to over-estimate capacities. Furthermore, these over predictions seem to be significantly influenced by eccentricity of loading. This dissertation presents a unified upper bound plastic limit analysis solution aiming to improve predictions of capacity for the aforementioned cases. In addition, a simplified upper bound method is proposed for cases in which computational efficiency is needed. Both solutions are compared to results from three dimensional finite element studies. Towards this end, most of the existing simplified predictive methods typically apply to idealized soil strength profiles that are either constant or linearly increasing with depth. However, site investigations often reveal complex strength profiles that deviate significantly from simple linear profiles. One example is the case in which a superficial stiff layer overlays a thicker layer of very soft soil. The work herein presented also includes analyses of pile and caisson performance in stratified soils based on a three dimensional upper bound PLA with a collapse mechanism comprising a surface failure wedge, a flow-around region and a spherical base failure surface. An introductory discussion on the influence of soil stratigraphy and geology for design purposes is included. Selected strength distributions are representative from field data obtained through cone penetration testing. Finally, the installation of driven piles and suction caissons in clayey soils generates excess pore pressures that temporarily reduce load capacity due to side resistance. Time dependent dissipation of these excess pore pressures leads to recovery of side resistance, a process known as 'setup'. Since many facilities cannot be put into operation until sufficient pile load capacity has been mobilized, realistic predictions of setup time can be important. A simplified method of analysis for calculation of the setup time following open ended pile penetration is also presented.
