The earthquake stability of earth structures
The main aim of this study is to develop a practicable computer method for analysis of the dynamic response of a two dimensional soil structure, which takes into account the generation of excess pore pressures and can predict final deformations. A secondary aim is to investigate the dynamic behaviour of a silt and the effect of variations in the stress-controlled dynamic triaxial test method on the measured sample response.
The soil model employed in the developed method of dynamic analysis uses a set of nested yield surfaces to represent shear stress-strain behaviour. These yield surfaces are defined by von Mises failure criterion, have an associated flow rule and undergo both kinematic and isotropic hardening. An empirical model is used to calculate the rise in pore pressure due to cyclic shear stresses. The changes in yield surface radii and hardening constants are defined by experimentally determined relationships between maximum shear modulus, shear strength and mean effective confining stress. A combination of critical state soil mechanics and volumetric considerations is used to relate increments in mean strain to increments in mean stress.
The soil structure is divided into elements and a finite difference scheme used to solve the equations of motion. An energy transmitting boundary is incorporated in the analysis. The program has the capacity to analyse the dynamic response of a soil structure to both shear and compression waves approaching the base of the structure from any angle.
The dam constructed for the Patea hydroelectric scheme was used as a practical application of the computer program. An input earthquake record considered appropriate to the site was chosen. The dynamic analysis of the dam with an empty reservoir indicated that in this state the dam can satisfactorily withstand the design earthquake. The dynamic analysis of the dam with a full reservoir subject to the same uniform base excitation suggests that a slide in the downstream shoulder would develop. The input earthquake waves travelling along the base of the dam from the upstream toe to the downstream toe, accentuate these displacements. These displacements are caused by the generation of high pore pressures which redistribute after the earthquake to temporarily reduce the stability of the dam further. The assumptions and approximations inherent in the method of analysis tend to make the results pessimistic. It was found that the vertical component of earthquake excitation was the major cause of the generated pore pressure.
The results of an extensive programme of laboratory tests carried out on the main construction material in the dam, are presented. The test programme consisted of undrained triaxial, isotropic consolidation, free vibration torsion and stress controlled dynamic triaxial tests. Factors influencing the pore pressure generation characteristics obtained from the stress-controlled dynamic triaxial tests, were found to include the compacted dry density, the cyclic shear stress, the initial effective confining stress, the test frequency, the method of sample compaction, the use of filter paper side drains and the type of end platen.