Walk‐away VSP using drill noise as a source

Abstract

We describe a method for extracting and deconvolving a signal generated by a drill bit and collected by an array of surface geophones. The drill‐noise signature is reduced to an effective impulse by means of a multichannel Wiener deconvolution technique, producing a walk‐away reverse vertical seismic profile (VSP) sampled almost continuously in depth. We show how the multichannel technique accounts for noise and for internal drill‐string reflections, automatically limiting the deconvolved data to frequencies containing significant energy. We have acquired and processed a data set from a well in Germany while drilling at a depth of almost 4000 m. The subsurface image derived from these data compares well with corresponding images from a 3-D surface seismic survey, a zero‐offset VSP survey, and a walk‐away VSP survey acquired using conventional wireline techniques. The effective bandwidth of the deconvolved drill‐noise data is comparable to the bandwidth of surface seismic data but significantly smaller than what can be achieved with wireline VSP techniques. Although the processing algorithm does not require the use of sensors mounted on the drill string, these sensors provide a very economic way to compress the data. The sensors on the drill string were also used for accurate timing of the deconvolved drill‐noise data.