Abstract
This paper presents the state of knowledge of analysis and design of rock sockets--main features of axial loading behavior and influential factors, methods for predicting the mobilized strength (shaft shearing f(s) and end bearing q(b)) and load-displacement (Q-w) curves. A total number of 363 field load tests on full-scale rock sockets are compiled into a database-DUT/PileROC/363 and analyzed in a consistent manner. This database is used to evaluate the variability of predictions of mobilized resistances (fs and qb), loads Q(a) at settlements w = 5-20 mm and 0.5%-2% of socket diameter, and Q-w curves. For fs, qb, and Qa, the results are characterized as the mean and coefficient of variation (COV) of the ratio of measured over predicted values. It is observed that (1) both shaft shearing and end bearing can be mobilized at small displacements when good construction and inspection procedures are followed, (2) the majority of empirical models for fs and qb are too simplistic (solely accounting for the unconfined compressive strength of rock) and the resulting predictions are highly dispersive (COV > 0.6), (3) the prediction quality does not necessarily improve with an increased level of model sophistication, (4) rock mass modulus Em is crucial for predicting Qa and Q-w curves, (5) predictions of Qa by one nonlinear and two side-slip design methods are of medium dispersion (COV = 0.3-0.6), and (6) only the nonlinear design method produces Q-w curves that resemble the shape of measured Q-w curves in load tests.