oa Petro-Elastic and Lithology-Fluid Inversion from Seismic Data, a Geophysical Perspective

Post-stack and pre-stack seismic inversion techniques are now widely used in the oil and gas industry for reservoir characterization. Nevertheless, significant challenges remain for a truly quantitative use of inversion results in constraining detailed reservoir models: seismic inversion is non-unique and usually unable to recover the fine vertical details required in geological models; the link between inverted elastic properties and petrophysical properties is also non-unique. Recent advances in seismic reservoir characterization combine rock physics and geostatistics in order to better constrain seismic and rock property inversions and quantify uncertainty. In this presentation, we first review seismicbased stochastic reservoir characterization workflows with examples from a giant, on-shore carbonate reservoir and from a deep water turbidite field offshore West Africa. In the carbonate example, we show how post-stack stochastic inversion is combined with stochastic porosity modeling to characterize the uncertainty in the spatial distribution of thin, low porosity intra-reservoir layers, which adversely affect the field water flood performance. In the turbidite example, we combine prestack stochastic inversion with Bayesian classification to derive detailed lithofacies-probability cubes and study the uncertainty in sand volume and well connectivity. In the second part of the presentation, we introduce the concept of direct petrophysical inversion, which involves a direct inversion of seismic amplitudes for rock properties such as porosity and saturation. We illustrate this technique using 3-D and 4-D inversion examples from the North Sea. Whether direct or cascaded inversion for rock properties is performed, Petro Elastic Models (PEM) play an increasingly important role in linking seismic and reservoir properties. In the presentation, we emphasize the role of statistical rock physics for incorporating uncertainty analysis with PEM transforms.