Controlled-source electromagnetic ͑CSEM͒ field surveys offer a geophysical method to discriminate between high and low hydrocarbon saturations in a potential reservoir. However, the same geological processes that create the possible hydrocarbon reservoir may also create topography and nearsurface variations of resistivity ͑e.g., shallow gas or hydrates͒ that can complicate the interpretation of CSEM data. In this paper, we discuss the interpretation of such data over a thrust belt prospect in deepwater Sabah, Malaysia. We show that detailed modeling of the key scenarios can help us understand the contributions of topography, near-surface hydrates, and possible hydrocarbons at reservoir depth. Complexity at the surface and at depth requires a 3D electromagnetic modeling code that can handle realistic ten-million-cell models. This has been achieved by using an iterative solver based on a multigrid preconditioner, finite-difference approach with frequency-dependent grid adaptation.

No abstract

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractThe need to rapidly produce a notional field development plan, to reduce costs and cycle-time has driven a fast-track reservoir characterization and reservoir model-building project for one of many fields offshore North West Borneo, Malaysia. The objectives of the study have been defined based on the deliverables stated by the reservoir engineers. Considering the list of data and limited time available, an efficient generalised inversion workflow was designed to process a large volume of deepwater 3D seismic data with a single exploration well to:1. Locate and map all sands (3D geobody identification) versus non-reservoir rocks. 2. Sub-divide reservoir sands into 4 lithofacies (well log neural net based calibration). 3. Propagate 4 lithofacies to entire reservoir volume (within two fluid types). 4. Integrate the fault framework with the sand and fluid distribution to build a first pass reservoir model. 5. Include multiple realizations to manage uncertainty. 6. Check sand connectivity and compute volumes of reserves in place. An innovative combination of 3D geostatistical and neural network techniques was used both for the well log data and for 3D seismic attributes (AVO, Acoustic impedance and dipazimuth combinations) to map the spatial distribution of the sand and their lithofacies. The results of the calibrated and quantitative generalised inversion were used in four different modes to assess the best way to build a first pass reservoir model and compute independent reserves within a short time-frame. This case study illustrates how a purposefully designed 3D/3D reservoir characterisation workflow can reduce the time required to build a first pass static reservoir model and how a similar process can be applied to other complex deepwater hydrocarbon accumulations. It focuses specifically on the different ways a static reservoir model can be built from 3D seismically derived volumes (3D/3D, Hybrid and grid-based).

No abstract

No abstract