ELSEVIER Soil Dynamics and EarthquakeEngineering 13 (1994) 139-145 © 1994 Elsevier Science Limited Printed in Great Britain. All fights reserved 0267-7261/94/$07.00 Substructure methods for impact systems: Comparison with free field measurements Th. Neidhart, J. Grabe & G. Huber Institute of Soil and Rock Mechanics, University of Karlsruhe, 76049 Karlsruhe, Germany Communicated by G. Borm (Received December 1991; revised version received November 1993; accepted 14 December 1993) An excitation system, consisting of a concrete foundation and a vibrator, was placed on a gravelly top layer over a 40 m deep soft clay deposit. The excitation frequency was varied and the acceleration of the foundation was measured simultaneously. Geophones were installed to monitor the velocities inside and on the surface of the subsoil. It was observed that the soil behaved nonlinearly and that the foundation lost contact with the soil at some excitation frequencies. A one mass-spring-dashpot system with the ability to consider the temporary loss of contact is used to model the movement of the excitation system. Inside the subsoil, the wave propagation is modeled using boundary elements in the time domain. Displacements and velocities are calculated at various points with this numerical tool. The results are compared with the measurements. INTRODUCTION In recent years settlements of historical buildings, especially those founded on soft clay layers, have been reported. It is to be assumed that resulting damages are partially caused by vibrations due to traffic and construction machinery in use near the buildings. Purely elastic methods are not suitable to describe such effects. A field test was carried out to investigate the influence of high strains on the soil-structure interaction and on wave propagation. An excitation system was placed on the gravelly top layer of a soft clay deposit in Konstanz, Germany. The excitation system was constructed to transmit a high amount of energy to the soft clay layer. The pore water pressure increased during the excitation, which indicates a high strain level. 1,2 The response of the system to a monochromatic excitation was not sinusodial. The system behaves nonlinearly because: • it temporarily loses contact with the soil and produces impacts, • high strains near the foundation reduce the low strain soil parameters and increase the pore water pressure. A procedure is introduced to model the loss of contact using piecewise linear systems. The nonlinear soil behaviour is approximated by using a decrease of the G-modulus depending on the strain level. 3'4 The calculation of the displacements and the velocities inside the deposit requires a numerical method. For that reason, some calculations with the boundary element method (BEM) in the time domain using fundamental fullspace solutions have been implemented. In recent years, the BEM has become a well known, established and often used numerical tool. The ability of this method is documented in Refs 5-11. A suitable numerical model for the current situation could be the FEM in combination with a practicable constitutive law for the near field connected with BEM. In contrast to such numerical calculations, the aim of our work was to check the ability of simpler methods to predict nonlinear dynamic in-situ effects. 139 THE FIELD TEST An open field with no external influences from neighbouring buildings and stiff underlying layers was chosen to perform the free field test. The top layer at the test field was mainly composed of sand and gravel. Below this first 2 m thick layer, a soft clay deposit with a thickness of up to 40m was found. The ground water table was rather constant. It remained during the whole year in the top layer between 0"2 and 0.5 m above the