The Learning from the Large Scale Lotung Soil-Structure Interaction Experiments

Blind prediction analyses and subsequent correlation studies of a 1/4-scale reinforced concrete containment model constructed at Lotung, Taiwan subject to forced vibration tests and actual earthquakes are evaluated with the objective of validating soil-structure interaction (SSI) analysis methodologies commonly used in U.S. practice. The SSI methods used range from simple soil-spring representation to more complex finite-element methods and sub structuring techniques. Both forced vibration test (FVT) data and actual earthquake induced response data have been obtained for use in validating selected SSI analysis methodologies. Considering that for forced vibration tests only the stiffness and damping characteristics of the foundation are required (complexities of site response, wave scattering and stiffness degradation of soils are absent), the FVT evaluation shows that acceptable frequency predictions can be obtained by most of the methods; however, soil damping as obtained from geophysical methods does not seem to account for the total energy dissipation during SSI. A number of insights have been obtained with respect to the validity of SSI analysis methodologies for earthquake response. Among these are the following: vertical wave propagation assumption in performing SSI is adequate to describe the wave field; equivalent linear analysis of soil response for SSI analysis, such as performed by the SHAKE code, provides acceptable results; a significant but non-permanent degradation of soil modulus occurs during earthquakes; the development of soil stiffness degradation and damping curves as a function of strain, based on geophysical and laboratory tests, requires improvement to reduce variability and uncertainty; backfill stiffness plays an important role in determining impedance functions and possibly input motions; scattering of ground motion due to embedment is an important element in performing SSI analysis; more than the calculational techniques, the differences in response predictions are due to the modeling of the soil-structure system