This project develops an innovative real-time simulation system of turbulent fluids, FlowBase, by employing a novel flow pattern database. A collection of pre-generated small flow patterns are managed by a database system with efficient storage and searching strategy, query and transaction handling, to enable fast and effective retrieval. The patterns are integrated with real-time ongoing simulation according to physical and topological parameters of the ongoing global flow. The FlowBase system is implemented with GPU (graphics hardware) acceleration on single PC or clusters. Furthermore, the system provides focus+context flow pattern integration that only appends turbulent flow details in focus areas, and also offers flexibility by adding patterns according to the computational resource budget. In this way, the system overcomes the resource limit for large-scale simulations and provides turbulent flow details for modeling visually-satisfying and physically-correct fluid phenomena.

The FlowBase system can rapidly provide vivid and spirited fluid behavior to create realistic experiences which are a decisive factor in many educational and entertainment applications, including virtual reality based education, advertisement and entertainment industry. It also benefits prediction and training applications, where fast computation and visual interaction of turbulent fluids are highly critical, such as contaminant dispersion prediction, emergency responder training, flight simulators, and urban and environmental design. The research outcome of the pattern database, novel algorithm, modeling techniques, computational and visualization tools are made publicly available through publications and software packages.

Project Report

Intellectual merits: This project develops an innovative simulation system of turbulent fluids. A collection of new methods are developed in the system to enhance the flow simulation technologies in different applications. The outcomes include: (1) adding turbulence to flow simulations through controllable external random forcing and novel Langevin Particles; (2) incorporating stochastic turbulence in particle-based fluid simulation and light-weight floating objects; (3) extracting flow patterns from topological abstraction of flow structures from ongoing flows; (4) proposing a novel pattern-based fluid animation platform for fluid control with patterns; (5) inventing effective compression, storage, management and efficient transfer of fluid simulation results. We also study interactive visualization system for large scale flow and data exploration. The system overcomes the resource limit for large-scale simulations and provides turbulent flow details for modeling visually-satisfying and physically-correct fluid phenomena. Broader impacts: The system rapidly provides vivid and spirited fluid behavior to create realistic flow simulation for education, entertainment and scientific applications. It also benefits prediction and training applications, where fast computation and visual interaction of turbulent fluids are very important. The critical applications include contaminant dispersion prediction, emergency responder training, flight simulators, and biomedical flow modeling for human health study. The research outcome of the new techniques, software system, computational and visualization tools are made publicly available through publications, software packages, websites and outreach activities.

Agency
National Science Foundation (NSF)
Institute
Division of Information and Intelligent Systems (IIS)
Application #
0916131
Program Officer
Jie Yang
Project Start
Project End
Budget Start
2009-07-15
Budget End
2013-06-30
Support Year
Fiscal Year
2009
Total Cost
$398,939
Indirect Cost
Name
Kent State University
Department
Type
DUNS #
City
Kent
State
OH
Country
United States
Zip Code
44242