Incorporation of seismic intermediate-scale data for improving reservoir modeling

A description of reservoir heterogeneities at different scales is crucial for improvements in reservoir characterization. In reservoir modeling approaches based on conventional seismic and log data the gap between short and large scales is usually filled by geological and/or geostatistical inferences, which are not guided by field data.; That lack of data providing intermediate-scale information is critical in fields where the distribution and spatial patterns of heterogeneities affecting important fluid flow properties are beyond the seismic lateral resolution and cannot be statistically inferred using the well distribution. Although being too expensive or impractical to “shoot” across every interwell distance, seismic profiles such as crosswell surveys can provide that information measured over small volumes.; In this work a multi-scale approach was developed to integrate seismic intermediate-scale data and the conventional short and large scale data, represented by logs and surface seismic data respectively. Intermediate-scale data obtained by short and far-offset crosswell surveys, vertical seismic profiles and high resolution seismic lines can be geostatiscally incorporated in this approach in order to provide more complete reservoir descriptions.; The approach employs the available seismic data as reference data in order to select traces of acoustic models derived from sequential Gaussian simulation honoring log data. At the end of an iterative process, a new high-resolution acoustic model is generated. The set of traces occupying interwell positions is filtered by 2D operators designed to mimic such specific crosswell imaging and each trace is then submitted to a post-matching operation using the collocated field trace.; The post-matching is designed to accept the simulated high resolution traces whose filtered versions have higher correlation with collocated field traces. These simulated high resolution traces serve as conditioning data for the next simulation. An interwell model at the intermediate scale is the final product of this first step. Such upscaling and post-matching is then repeated to the larger-scale seismic traces, using logs and the interwell image as hard data to feed simulations. The final product is an optimal acoustic property model constrained by all logs and by seismic surveys.; This multi-scale incorporation was applied using field and synthetic data. From the synthetic study, I found that the main impact on the model is caused by the incorporation of crosswell images obtained by shooting short distance surveys. Short-offset surveys provide a tool to map and recover a very fine layered and lateral framework. Crosswells and logs can be used to stochastically draw interwell models, which are then extrapolated under the constraint of conventional seismic data.; The field study using data from McElroy Oil Field, West Texas, showed that only the short-offset crosswell survey (interwell distance of 187 feet) can provide very high-resolution interwell image containing the necessary geostatistical link between acoustic properties measured at logs and at log-collocated traces. Higher vertical resolution tools like log data and short-offset crosswell image can be used to reduce large-scale seismic uncertainties providing improvements in reservoir acoustic models.; Since some reservoir and acoustic properties are directly related, the reduction of uncertainty achieved by building these acoustic models can be transferred to models of volumetric properties such as porosity. The integration must prioritize data with higher frequency content, i.e., should follow two steps: only short-crosswell data are used in the first post-matching, and additional lower resolution seismic data during the second.