3D visualisation of oil reservoirs

Mulholland, Samantha (2017) 3D visualisation of oil reservoirs. MPhil(R) thesis, University of Glasgow.

Abstract

This research introduces a novel approach to storing compressed 3D grid information by applying octree compression techniques. This new data structure stores the octree in a pruned flattened fashion where only header and active leaf nodes are stored in a linear array. This generates high levels of lossless compression when applied to 3D geometry where clusters of homogeneous information exist. This data structure yields fast, log(n) look-up times and initial results show that when coupled with bespoke scanning methods searching times can surpass that of direct access. Hierarchical pyramid visualisations techniques are also presented using the information stored at each level in the tree structure. Integrating with this are face culling algorithms developed in this research, which eliminate hidden face and inner leaf node cells which eases the burden placed on the CPU and GPU. By integrating these pyramid scaling and face culling algorithms, grid models can be shown at various levels of resolution incorporating sub-regions, "regions of interest" displayed at full resolution. This further lightens the load on the GPU generating quicker loading times and higher refresh rates. This can potentially allow larger models to be visualised than would otherwise have been possible.

This research was sponsored by Sciencesoft an oil reservoir visualisation company and the algorithms developed in this research have been applied to compressing oil reservoir information. Oil companies require accurate 3D computer-generated models of oil reservoirs in order to make oil and gas extraction as cost effective as possible. Advances in computing power has meant that it is now possible to run multi-million cell oil reservoir grid models, increasing the level of accuracy and precision available to engineers. This thesis applies 3D octree compression techniques to these computer models and compares these with industry standard storage and cell searching algorithms as industry benchmarks. This thesis suggests that octree compression techniques may prove to be a more efficient data structure for storing and searching active cell information within oil reservoirs than existing procedures.

• Item Type: Thesis (MPhil(R))
• Qualification Level: Masters
• Keywords: 3D visualisation, Octree compression.
• Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
• Colleges/Schools: College of Science and Engineering > School of Computing Science
• Supervisor's Name: Cockshott, Dr. Paul
• Date of Award: 2017
• Depositing User: Ms Samantha Mulholland
• Unique ID: glathesis:2017-8590
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 07 Nov 2017 13:28
• Last Modified: 25 Jul 2018 13:03
• URI: https://theses.gla.ac.uk/id/eprint/8590