Advances in 4D Seismic: Application in the North Sea

摘要

This paper describes an integrated approach to 4D seismic, using legacy time-lapse 3D data in the UK North Sea. The baseline 3D surveys are typically 1980s' vintage, with time-lapse 3D being acquired after 10 years. The time-lapse surveys have been acquired for structural definition and are therefore only fortuitously 4D. After re-gridding and re-aligning the 3D traes to the same spatial location for each vintage, both post-stack and AVO inversion is applied, using the same wells to derive both low frequency models and wavelets in each vintage. This inversion calibrates the seismic in acoustic impedance units relative to the well logs, after removing the (different) wavelet signatures from each vintage. The resulting acoustic impedance volumes are subtracted one from another to provide difference cubes. The seismic analysis is complemented by forward modelling from the reservoir simulator model, i.e. to a range of synthetic acoustic impedance volumes. The synthetic volumes are defined by the spectrum of responses 'allowed' in the current field development scenarios and the comparison with real data tests the reservoir model. In addition, forward modelling into the future predicts the best dates for acquisition of new seismic data. A parallel approach conditions assessments of history matching to the observed seismic response. The method finds application in field development planning, and monitoring of producing reservoirs (oil and gas). History matching and reservoir models are updated with reference to the 4D seismic data. Poorly depleted fault blocks are identified, and flow patterns deduced in water-injection. Pressure variations across fields, including unexpected gas evolution are observed. The paper suggests that 4D seismic can improve development planning and reduce costs over the field lifetime. 4D seismic identifies un-swept oil and improves the recovery factor. We also demonstrate that legacy data-sets are important information for field monitoring.