Troy Thompson

A study of stress induced anisotropy using the 'relaxation' method on synthetic core

Summary The seismic effects of different stresses on rock core are commonly studied using whole core under different amounts of pressure. Normally this requires positioning whole core within a total jacket which exerts triaxial pressure on it while ultrasonic transducers transmit both compressional and shear energy across the core uniaxially. The result is a better understanding of stress fracturing as a function of the application of the stress. Such a study requires a major test facility. We present an alternative approach to assist the understanding of the effects of stress, by forming synthetic whole core under pressure, and after relaxation, studying the effects of seismic wave propagation on the relaxationinduced fracture system. While this cost-effective method requires little more than a car jack and ultrasonic transducers, it can be extended to provide useful knowledge of the effects of other matrix parameters.

Integration of uncertain subsurface information into multiple reservoir simulation models

requires the integration of uncertain subsurface information into multiple reservoir simulation models. This information includes seismic data, various types of well data, and geologic concepts. Over the past five years, a workflow has been developed by various organizations in conjunction with BHP Billiton Petroleum. This distinctive approach focuses first on building mesoscale reservoir models that can be constrained by seismic data (typically with a resolution up to the stratigraphic seismic loop scale, see Prather et al., 2000), then introducing the finer scale geologic concepts and well data needed for reservoir simulation models (stratigraphic firstand second-order subseismic scale, where each order is about a factor of three in size) via a downscaling step that honors mesoscale model constraints. Uncertainty and correlations of the well and seismic measurements are always taken into account. In fact, they are necessary to be able to combine the various measurements. Bayesian probabilistic techniques are used extensively in this process. The result is an ensemble of reservoir simulation models consistent with all subsurface information. The application to Stybarrow Field, in the Carnarvon Basin of offshore Western Australia, will be used as an example of this workflow. This workflow starts with a conventional correlated wavelet extraction and sparse spike inversion. The sparse spike inversion gives a preliminary estimate of net rock volume and fluid probabilities. Although it does not consider uncertainty, multiple layer seismic interference or many well constraints, it helps build a layer-based model framework for subsequent steps where these influences are considered. Because we will be estimating the uncertainty, we need to know what the uncertainty of the seismic data is. To obtain this, the wavelet derivation needs to produce an estimate of the seismic noise—that is, the part of the seismic data that does not correlate with the synthetic of the well log. We used a probabilistic wavelet derivation based on Bayesian concepts that predicts the noise level. Given the wavelet with uncertainty, we then perform a trace-based probabilistic model-based inversion which combines the seismic data with well measurement constraints to give average values, uncertainties, and correlations in important properties such as net reservoir sand, net-to-gross sand, and fluid content. Several significant technical challenges remain. First, the results must be “massaged” onto the reservoir simulation grid (which is commonly nonuniform in space). Second, transverse spatial correlations that are geologically realistic must be enforced, while eliminating short scale fluctuations that are artifacts of the trace-based inversion. Geostatistical techniques are used to introduce this lateral correlation while respecting the nontrivial correlations produced by the model-based inversion. The final challenge is to “decorate” the model with the stratigraphic firstand second-order subseismic structure needed to allow the reservoir simulation to assess the flow effects of subseismic heterogeneity. The subseismic structure is based on geologic concepts and is constrained by well and seismic information. An “enforcement” step then slightly deforms the decorated model to match the gross thickness and net sand estimates from the inversion. Results of all steps will be shown for Stybarrow Field. Integration of uncertain subsurface information into multiple reservoir simulation models

Prestack hyperspace propagation for automated event picking

Summary Seismic data mining is part of an interactive processing and interpretation workflow. The extraction of information will often have the prerequisite of picking reflection events. Methods that aid in automatically extracting information are required when handling large volumes of data. Migrated 3-D seismic data in prestack form (which includes the offset dimension) creates a 4-D hyperspace. An algorithm for tracking prestack reflection events in that hyperspace will be presented. The algorithm combines a range of techniques including supervised learning. Results of automated picking will be presented for migrated, prestack, field 3-D data. The algorithm was able to track a nominated reflection event in prestack hyperspace from a single seed pick. The results are superior to those produced using a 2-D gather-based approach and a correlation autopicker.