This case study presents the impact of dynamically updating 3D structural and petrophysical geological models to optimize real-time geosteering through stringer sands with high effective porosities in clastic reservoirs. The depositional environment of siliciclastic reservoirs imposes reservoir development challenges due to heterogeneity within the play system and within each sediment. Historically, stochastic modeling techniques have been used to predict the extent of lithofacies between wells where no hard data exists. Combined with a proper understanding of the depositional environment, stochastic and deterministic modeling techniques (selection of which depends on the availability of well control data) can closely simulate the actual distribution of petrophysical properties and their variance within reservoirs.

Reservoir modeling has been used for reservoir studies and production simulation, historically with cell resolutions as coarse as 100's of meters/feet. Such models used to require hours to update. Today, advancements in modeling software and hardware allow higher resolution models to be updated in minutes. These advancements have triggered new opportunities for modeling, allowing model-centric approaches to be effectively integrated into time-sensitive workflows. Geosteering is one of the evolving reservoir development and drilling frontiers where accurate real-time decisions are required to achieve drilling objectives with minimum cost.

This paper demonstrates a new approach to updating an integrated reservoir model in near real-time, both structurally and petrophysically, to optimally geosteer wells through the most productive zones of the reservoir. The proposed workflow assumes the availability of pre-built geological models populated with at least a porosity attribute. Instead of relying on quick-look effective porosity logs, which are subject to many unknown parameters, we present a new methodology to convert porosity models to gamma ray models which can be more easily correlated with the LWD gamma ray log. This minimizes the number of computational steps required to update the model, resulting in faster model updates.

To further accelerate model update, models were sectored (cropped) around the active well. Sector boundaries need to be carefully selected to honor the influence of adjacent wells during krigging. Smaller models allow increased horizontal and vertical resolution, to approximate the well log resolution, without impacting compute time. This high resolution model is then populated with the full field petrophysical properties and used for geosteering. The paper concludes by summarizing lessons learned using this approach, the pros and cons, and possible areas for future enhancement.

This content is only available via PDF.
You can access this article if you purchase or spend a download.