Visualization of Dynamic Uncertainties.

Thomas Viard and Emmanuel Gringarten and Alexandre Hugot and Guillaume Caumon and Bruno Levy. ( 2010 )
in: Proc. 30th Gocad Meeting, Nancy

Abstract

Understanding spatial uncertainty is critical for the assessment of a reservoir, as it strongly impacts decision-making in reservoir development, e.g. for the placement of infill wells. 3D uncertainty visualization can provide clear and intuitive insights on subsurface property and the associated spatial uncertainty. Most available methods however solely consider static uncertainties and disregard time-dependent uncertainties related to dynamic properties (e.g. oil saturation in a flow simulation). We propose a new approach which combines static uncertainty visualization methods with time-dependent animation schemes, in order to visualize uncertainty through the time-steps of several flow simulations or time-lapse seismic characterization. In order to avoid distracting lags, it is crucial to ensure that images are produced at interactive framerates (at least 20 images per second). We therefore exploit the characteristics of modern graphic cards (GPUs) to achieve efficiency even on large datasets: we store the time-dependent properties and uncertainty fields into textures for fast data access, and we blend the properties directly on the GPU to benefit from the GPUs’ massively parallel architecture. Our approach is illustrated on the Brugge dataset with thirty different flow simulations. It clearly demonstrates the effects of passing faults on water saturation uncertainty, which can be useful in order to select some locations were new production wells could be drilled.

Download / Links

BibTeX Reference

@inproceedings{Viard2GM2010,
 abstract = { Understanding spatial uncertainty is critical for the assessment of a reservoir, as it strongly impacts decision-making in reservoir development, e.g. for the placement of infill wells. 3D uncertainty visualization can provide clear and intuitive insights on subsurface property and the associated spatial uncertainty. Most available methods however solely consider static uncertainties and disregard time-dependent uncertainties related to dynamic properties (e.g. oil saturation in a flow simulation). We propose a new approach which combines static uncertainty visualization methods with time-dependent animation schemes, in order to visualize uncertainty through the time-steps of several flow simulations or time-lapse seismic characterization.
In order to avoid distracting lags, it is crucial to ensure that images are produced at interactive framerates (at least 20 images per second). We therefore exploit the characteristics of modern graphic cards (GPUs) to achieve efficiency even on large datasets:
we store the time-dependent properties and uncertainty fields into textures for fast data access, and we blend the properties directly on the GPU to benefit from the GPUs’ massively parallel architecture. Our approach is illustrated on the Brugge dataset with thirty different flow simulations. It clearly demonstrates the effects of passing faults on water saturation uncertainty, which can be useful in order to select some locations were new production wells could be drilled. },
 author = { Viard, Thomas AND Gringarten, Emmanuel AND Hugot, Alexandre AND Caumon, Guillaume AND Levy, Bruno },
 booktitle = { Proc. 30th Gocad Meeting, Nancy },
 title = { Visualization of Dynamic Uncertainties. },
 year = { 2010 }
}