Summary
Nine wells were drilled to test cyclic steam stimulation as a recovery mechanism in the diatomite reservoir in the Belridge field. Micro-seismic monitoring was proposed to evaluate steam-chest and fracture growth. A series of models were constructed to determine both microseismic-event detectability and locatability. The modeling indicated that poor signal/noise ratios would constrain the ability to locate events using a single array. As a result, three microseismic-monitoring wells (MOWs) were installed. The purpose of this paper is to evaluate the microseismic-event-location results obtained from the three-well solution and compare them with the solutions obtained when turning off one or two of the arrays. This first phase of investigation was performed on the sand-propped hydraulic-fracture stimulation before the cyclic steam operations. This study may be applied to other areas, including imaging hydraulic-fracture stimulations in shale plays, reservoir steam monitoring, or in any area where location precision in microseismic monitoring is necessary.
It has been demonstrated in the field that the monitoring range of a microseismic system can be increased and the potential for locating microseismic events improved by installing multiple arrays having overlapping radii of observation. Innovative system design, deployment techniques, operational procedures, and advanced multiwell processing strategies have all contributed to creating a growing data set with more than 4,000 events recorded during the first 13 months of operation. Results of turning off arrays showed increased location error with two-well solutions and a significant increase in error with one-well solutions, notwithstanding the reduction in event location count because of a lack of multiphase signals on a single array. Azimuth errors in single-microseismic-observation-well solutions result in dispersed interpreted fracture geometry and in misinterpretation.
Overall, our observations show that for both multi-and singlephase events, the observed detection limits and size distribution of the seisms have far exceeded those originally predicted by the earlier modeling.