Dynamic Updating of Reservoir Models

Abstract

This article, written by Senior Technology Editor Dennis Denney, contains highlights of paper SPE 123671, ’Dynamic Updating of Reservoir Models,’ by Harun Ates and Asnul Bahar, SPE, Kelkar and Associates, and Vitaly Krasnov, SPE, Rosneft, and Mohan Kelkar, SPE, University of Tulsa, prepared for the 2009 SPE Annual Technical Conference and Exhibition, New Orleans, 4-7 October. The paper has not been peer reviewed. In sandstone reservoirs, an important challenge is scaling up a fine-scale model effectively. Conventional methods are not adequate when significant shale distribution exists in the reservoir. In the fine-scale model, some sand bodies are connected to the wells and some are not. As the fine-scale model is scaled up, some discontinuous sand bodies become combined with connected sands. A new procedure was developed to overcome this problem. Introduction Traditional reservoir modeling involves developing a fine-scale geocellular model that incorporates the small-scale static uncertainties. Once such a description is created, it is scaled up to an appropriate scale such that it can be flow simulated to understand the dynamic performance of the reservoir. It is difficult to flow simulate a fine-scale model because it is computationally demanding. Some sandstone reservoirs tend to be highly discontinuous, with sand and shale dispersed within the reservoir. The continuity of the sand bodies may not be known precisely and reflects an uncertainty in static models. The problem becomes even more complex if the reservoir is thick and contains multiple thin sand bodies. When a fine-scale model is constructed, it may contain many vertical layers to account for all the thin sands present in the reservoir. Scaling up of such models is very difficult because of the following. If the model is coarsened (scaled up) significantly, many of the discontinuous sands will connect to other sands, and this will increase the pore volume (PV) “connected” to the wellbore, changing the transmissibility (connectivity) distribution between wells. In the scaled-up model, because of artificially connected sands, infill wells would not indicate true potential of additional recovery because some of the virgin sands that would have been connected to the newly drilled wells would show to have been drained already by existing wells in the model. If the model is maintained at relatively fine scale, the simulation can be computationally demanding. The dynamic modeling (e.g., history matching) would be more time consuming. Therefore, many uncertainties may not be investigated correctly.