Abstract:
The Downhole Seismic Test (DST) is an extensively utilized geotechnical tool for site characterization. DST provides low strain (< 10-5) in-situ interval shear and compression wave velocity estimates. These velocities are determined by obtaining relative arrival times of source waves as they travel through the stratigraphy and are recorded by one or more vertically offset seismic sensors. The DST arrival times have associated measurements errors and resolution limitations which become more pronounced as the depth interval of analysis is reduced. The errors in arrival time measurements result in extensive fluctuations in the estimated interval velocities with numerous outliers. BCE has developed a new DST analysis technique where analytically modelling of the DST arrival time data sets is accomplished by fitting high order polynomials. The main advantages of this new technique are five-fold. 1) Ability to utilize all arrival time estimates irrespective of measurement errors. 2) Ability to process small depth interval (≤ 0.5m) arrival time data sets. 3) Analytical polynomial “ best fit” function allows for user specification of desired depth intervals for data interpolation. 4) Facilitates sophisticated data fusion for significantly more accurate DST interval velocity estimation. 5) RMS error residual quantifies how well the “ best fit” polynomial fits the acquired arrival time data sets. This paper outlines the mathematically details of the 7th order best fit polynomial where a Kalman filter formulation is implemented. The analysis of several onshore and offshore real data DST data sets is also presented in this paper.