4D Time-Lapse Studies and Reservoir Simulation to Seismic Modeling

Abstract

Introduction 4D (time-lapse) seismic methods have made an important contribution to field development for several years in the North Sea and are increasingly being applied in other oil and gas provinces. Many studies relate qualitative relationships between seismic amplitude variations and changes in saturations and pressures in producing fields and an early development was the recognition of the importance of relating the observed changes to changes predicted by dynamic modeling (Watts et.al.(1996)). Recent applications of the method have emphasized quantification of reservoir rock property changes from 4D observations, and the importance of updating the dynamic model as a consequence.(Boyd-Gorst et.al. (2001), McInally et al. (2003)). By 'simulation-to-seismic' studies we mean forward modeling from the dynamic subsurface model to synthetic geophysical data and consequent comparison between the predicted and actual geophysical observations. The reverse path, seismic-to-simulation, is a classical inversion problem, where observations of the seismic changes are inverted against a dynamic model. The objective of all of these types of study is to target un-swept and near-field HC's and reduce uncertainties in predicting 'left-behind' volumes and increase the net recovery factors. Under this type of analysis seismic, reservoir and petroleum engineering studies are linked in a circular work-flow, with the emphasis being placed on the interface between static and dynamic models, 4D seismic and rock physics. Recent enhancements of the method include vertical ray-tracing through the simulation model, to determine rock properties at the seismic sample interval, both spatially and in time, comparison of flow-paths deduced directly from the seismic with those from the simulation, and more quantitative voxel-to-voxel comparisons of the predicted and observed responses. Along the way we have to address resolution differences between seismic and dynamically modeled data, adding to the complications posed by those associated with the differences between static and dynamic models. Subsurface properties are sought that bridge the divide between reservoir engineering and geophysics, e.g.. acoustic impedance, Poisson's Ratio, density, synthetic and real seismic trace information. Streamline simulators and streamlines generated from gradients of rock physics quantities derived from inversion suggest a natural method for comparing simulator and seismic results as a supplement to the voxel sample based comparison. The techniques are being applied to a range of geological settings, and for both oil and gas fields. Early indications for full 4D inversion (i.e. from seismic observations to applied to dynamic model), using simultaneous algorithms based on the current capabilities with AVO and 4D 'conventional' inversion methods, are very encouraging. Figure 1 A Circular Workflow for Integration of 4D Seismic and Reservoir Engineering :(Available in full paper) Simulation-toSeismic Methods: Figure 2 The Simulation-to-Seismic Workflow(Available in full paper) The input parameters (top of figure 2) that describe the model for dynamic simulation, have to be resampled and converted from depth to time to provide synthetic seismic responses. The most compute-efficient approach is to carry out vertical ray-tracing for each seismic trace through the model, and in general several seismic samples (4ms) are required for each reservoir block. Antialiasing filters are required in the vertical sense to eliminate potential problems with under-sampled thin units in the simulator.