A texture-based framework for improving CFD data visualization in a virtual environment

Bivins, Gerrick
Journal Title
Journal ISSN
Volume Title
Research Projects
Organizational Units
Journal Issue

In the field of computational fluid dynamics (CFD) accurate representations of fluid phenomena can be simulated but require large amounts of data to represent the flow domain. Inefficient handling and access of the data at initialization and runtime can limit the ability of the engineering to quickly visualize and investigate the entire flow simulation, and thus hampering the ability to make a quality engineering decision in a timely manner. This problem is amplified "n-fold" if the solution set is time dependent, or transient. To visualize the data efficiently, dataset access should be decreased if not eliminated at runtime to provide an interactive environment to the end user. Also a reduction in the size of the initial datasets should be reduced as much as possible while maintaining validity of the solution so that larger (i.e. transient) solution datasets can be visualized. To accomplish this, the format in which the dataset is stored should be changed from conventional formats. With the recent advancements of graphical processor unit (GPU) technology, current research in the computer graphics community has lead a novel approach for efficiently storing and accessing flow field data as texture data during a visualization. A so-called "texture-based" solution for visualization of flow fields allows the end user to visualize complex three-dimensional flow fields in an intuitive fashion while remaining interactive. This work presents a framework for incorporating texture-based analysis techniques into a current CFD visualization application to improve the capabilities for investigating flow fields. The framework presented easily extendible to allow for research and incorporation of progressive visualization methods, in keeping with current technology. Comparisons of the current framework with the texture-based framework are shown to effectively visualize a dataset that could not be visualized in its entirety with the current framework. Comparisons of common visualization techniques, such as contour planes and streamlines, are made to show how the texture-based framework out performs the current framework.

Mechanical Engineering