Estimation of soil shear modulus softening during strong ground shaking using ground surface and downhole acceleration recording

Technical Abstract

A growing number of earthquakes have occurred close to downhole accelerometer arrays. The resulting acceleration records from different depths provide valuable insight into the propagation and amplification of seismic waves in the site soils. Interpretation of data from these events frequently involves estimation of shear modulus for the soils involved. While estimates for small strain elastic deformations are usually available, softening due to nonlinear soil behaviour introduces unwanted complications in any estimation of shear moduli during strong shaking. A commonly used approach is to model the soil column as a one-dimensional shear beam and adjust the soil parameters throughout the period of shaking to best fit the observed data. The method is both awkward and unreliable since the shear beam model presupposes the form of response for the soil system. An improved method that directly uses the cross-correlation of acceleration records is proposed here. The research proposed in this report develops an algorithm for estimating the average shear modulus directly from measured acceleration data at different depths. The algorithm relies on cross-correlation of two records together with time compression of one of the records to automatically estimate the average shear wave velocity as a continuous function of time while shaking progresses.