Calibration and use of the petroelastic model for 4D seismic interpretation

Abstract

One of the major objectives of 4D studies is to understand and quantify changes in the seismic response, related to pressure and saturation changes in the reservoir. A key ingredient of such interpretation is the petroelastic model (PEM), which links fluid saturations and pore pressure changes in the reservoir rock to the elastic property changes required for seismic modelling, time-lapse feasibility studies, 4D inversion and also seismic history matching. Many previous studies have pointed out the difficulty of selecting a PEM, the challenges in calibrating the model to the in situ response, and in particular the uncertainties involved. In this work I study the use of different deterministic PEMs for simulator to seismic modelling. The models are applied to three fields in the UKCS, Norwegian Sea and offshore Brazil with distinctly different geological settings. For each model, the static components are calibrated against a range of wireline log data acquired prior to production using an optimization algorithm. The dependence on pressure change is then added separately using coefficients derived from core data in the laboratory. All PEMs for the clastic datasets as well for the carbonate reservoir, are found to yield similar responses making the choice of the “best” challenging; however even when an appropriate well calibrated model is used, it may not be adequate for time-lapse seismic studies. In addition, the large number of input parameters for each model makes the process of model fitting particularly non-unique. To reduce the input parameters for computing the time lapse seismic response a linear two parameter equation has been suggested which gives similar results to the multi-parameter models, specifically for an oil-water system. When working with maps of 4D seismic attributes in particular, the work presented in this research advocate that a simple model which is linear with respect to the pressure and saturation changes, and is “primed” by any conventional deterministic PEM may be an adequate alternative for time-lapse seismic interpretation. The non-unique nature of the rock physics models, together with data and model uncertainties creates the need for time-consuming comparisons in the Seismic History Matching (SHM) workflow. This study presents a simple and interactive way of visualizing all of these uncertainties, whilst optimizing the SHM. It consists of a simple cross-plot of all changes in water saturation and pore pressure between at least two time periods of interest in the reservoir history (usually pre-production baseline and a monitor) from the simulation model and colour coded based with the 4D seismic signature from the “predicted” models or “observed” data. Therefore it allow us to discriminate between regions in the reservoir that are dominated by pressure or saturation, from which a boundary line associated with the controlling parameters (CP/CS) from the simplified proxy model PEM can be drawn. Application to synthetic examples and field datasets verified the usefulness of this approach and the sensitivities to both data and the model.