Semi-implicit representation of discrete black oil reservoir models

Abstract

Numerical black oil reservoir simulation is widely employed in the oil and gas industry for planning and predicting field explorations. Reservoir simulation relies on a topologically structured grid of hexahedral cells, modeling complex geological features. Visualization of these large-scale simulated discrete models is essential for inspection and analysis but presents challenges due to their size and geometric complexity. While conventional methods focus on rendering external faces, advanced techniques, like volume rendering, require full model representation on the GPU and an efficient point-location algorithm. Previous works proposed GPU-based model representations, but real-time applications remain limited by performance. Simplified alternatives improve speed but at the cost of accuracy. This work introduces a compact semi-implicit representation of discrete reservoir models that combines a topological coordinate field and a depth map to enable the implementation of fast and accurate visualization algorithms. Computational experiments demonstrate the advantage of the proposed representation regarding memory usage and performance while preserving the visual quality of rendered images.