TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractThe fluids in the reservoir are redistributed in response to pressure gradients caused by hydrocarbon production, which is often coupled with injection of water or gas. It is known from density logging that the density of pore fluids is an important physical property useful in inferring oil, water, or gas saturation in a rock. Thanks to the density difference of the different phases, the borehole gravity measurement is a candidate for tracking the movement of fluids hundreds of feet away from wellbores. Such data are measured in time lapse to provide the three-dimensional distribution of density changes in time through an inversion procedure.Because gravity is a potential field, the inversion of borehole gravity data is inherently non-unique. The decrease of the sensitivity to the recorded data away from the measurement location is also a challenge for the inversion problem. The approach implemented in this work uses an iterative algorithm that minimizes a global objective function. The objective function includes the data misfit functional and a three-dimensional regularization that is needed to constrain the inversion to reasonable solutions. A weighting function based on the distance between each gravity source and the recording instrument is introduced to help mitigate the geometric decay of gravity kernels with distance from the sensor. Synthetic inversions are presented that indicate borehole gravity can be instrumental in detecting and monitoring a large water front. This study also points out the optimal conditions for using borehole gravity for fluid front monitoring, in terms of number of wells, relative well positions to the target, vertical sampling, and measurement accuracy.

No abstract