Load and resistance factor design (LRFD) (often called limit states design (LSD)) has been mandated in the AASHTO Bridge Design Specifications and will be adopted in future editions of Canadian Highway Bridge Design Code for all transportation-related structures including reinforced soil retaining walls. The ultimate objective of this thesis work was to carry out reliability-based analysis for load and resistance factor design calibration for rupture and pullout limit states for steel and geosynthetic reinforced soil walls under self-weight and permanent surcharge loading conditions. In order to meet this objective it was necessary to generate large databases of measured load and resistance data from many sources and in some cases to propose new design models that improve the accuracy of underlying deterministic load and resistance models. Numerical models were also developed to model reinforced soil wall performance. These models were used to investigate load prediction accuracy of current analytical reinforcement load models. An important feature of the calibration method adopted in this study is the use of bias statistics to account for prediction accuracy of the underlying deterministic models for load and resistance calculations, random variability in input parameter values, spatial variation and quality of data. In this thesis, bias is defined as the ratio of measured to predicted value. The most important end product of the work described in this thesis is tabulated resistance factors for rupture and pullout limit states for the internal stability of steel and geosynthetic reinforced soil walls. These factors are developed for geosynthetic reinforced soil wall design using the current AASHTO Simplified Method, a new modified Simplified Method, and the recently proposed K-Stiffness Method. Useful quantitative comparisons are made between these three methods by introducing the concept of computed operational factors of safety. This allows designers to quantify the actual margin of safety using different design approaches. The thesis format is paper-based. Ten of the chapters are comprised of journal papers that have been published (2), are in press (2), in review (3) and the remaining (3) to be submitted once the earlier background papers are accepted. / Thesis (Ph.D, Civil Engineering) -- Queen's University, 2010-01-28 18:07:22.284

Use of soil retaining walls for roads, embankments and bridges is increasing with time and reinforced soil retaining walls are found to be very efficient even under critical conditions compared to unreinforced walls. They offer competitive solutions to earth retaining problems associated with less space and more loads posed by tremendous growth in infrastructure, in addition to the advantages in ease and cost of construction compared to conventional retaining wall systems. The study of seismic performance of reinforced soil retaining walls is receiving much attention in the light of lessons learned from past failures of conventional retaining walls. Laboratory model studies on these walls under controlled seismic loading conditions help to understand better how these walls actually behave during earthquakes. The objective of the present study is to investigate the seismic response of geosynthetic reinforced soil wall models through shaking table tests. To achieve this, wrap faced and rigid faced reinforced soil retaining walls of size 750 × 500 mm in plan and 600 mm height are built in rigid and flexible containers and tested under controlled dynamic conditions using a uni-axial shaking table. The effects of frequency and acceleration of the base motion, surcharge pressure on the crest, number of reinforcing layers, container boundary, wall structure and reinforcement layout on the seismic performance of the retaining walls are studied through systematic series of shaking table tests. Results are analyzed to understand the effect of each of the considered parameters on the face displacements, acceleration amplifications and soil pressures on facing at different elevations of the walls. A numerical model is developed to simulate the shaking table tests on wrap faced reinforced soil walls using a computer program FLAC (Fast Lagrangian Analysis of Continua). The experimental data are used to validate the numerical model and parametric studies are carried out on 6 m height full-scale wall using this model. Thus, the study deals with the shaking table tests, dynamic response of reinforced walls and their numerical simulation. The thesis presents detailed description of various features and various parts of the shaking table facility along with the instrumentation and model containers. Methodology adopted for the construction of reinforced soil model walls and testing procedures are briefly described. Scaling and stability issues related to the model wall size and reinforcement strength are also discussed. From the study, it is observed that the displacements are decreasing with the increase in relative density of backfill, increase in surcharge pressure and increase in number of reinforcing layers; In general, accelerations are amplified to the most at the top of the wall; Behaviour of model walls is sensitive to model container boundary. The frequency content is very important parameter affecting the model response. Further, it is noticed that the face displacements are significantly affected by all of the above parameters, while the accelerations are less sensitive to reinforcement parameters. Even very low strength geonet and geotextile are able to reduce the displacements by 75% compared to unreinforced wall. The strain levels in the reinforcing elements are observed to be very low, in the order of ±150 micro strains. A random dynamic event is also used in one of the model tests and the resulted accelerations and displacements are presented. Numerical parametric studies provided important insight into the behaviour of wrap faced walls under various seismic loading conditions and variation in physical parameters.