Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


thesis.pdf (20.9 MB)

ETD Abstract Container

Abstract Header

Data Summarization for Large Time-varying Flow Visualization and Analysis

Chen, Chun-Ming
http://orcid.org/0000-0002-1662-1026
http://rave.ohiolink.edu/etdc/view?acc_num=osu1469141137

Abstract Details

2016, Doctor of Philosophy, Ohio State University, Computer Science and Engineering.
The rapid growth of computing power has expedited scientific simulations which can now generate data in unprecedentedly high quality and quantity. However, this advancement has not been mirrored in I/O performance, and hence scientific research is facing great challenges in visualizing and analyzing large-scale simulation results. Among areas of scientific research, fluid flow analysis plays an important role in many disciplines such as aerospace, climate modeling and medicine applications. The data-intensive computation required for fluid flow visualization makes it difficult to devise efficient algorithms and frameworks for flow analysis. First, to analyze a time-varying flow field, pathline visualization is typically used to reveal particle trajectories in the flow. Pathline computation, however, has irregular data access pattern that complicates out-of-core computation when the flow data are too large to fit in the main memory. Strategies on modeling the access pattern and improving spatial and temporal data locality are needed. Second, to avoid tremendous I/O latency, the simulated flow field results are typically down-sampled when they are stored, which inevitably affects the accuracy of the derived pathlines. Error reduction and modeling becomes important to enable uncertainty visualization in order for better decision making. This dissertation addresses the above challenges by data summarization approaches that efficiently process large data into succinct representations to facilitate flow analysis and visualization. First, a graph modeling approach is employed to encode the data access pattern of pathline computation, with which a cache-oblivious file layout algorithm and a work scheduling algorithm are proposed to optimize disk caching during out-of-core pathline visualization. Second, an incremental algorithm is devised that fits streaming time series of flow fields into higher-order polynomials and estimates errors in a compact distribution model. The benefit of this distribution-based error modeling is demonstrated to enable probabilistic uncertain pathline computation. Finally, a case study of jet engine stall is conducted for large flow simulations. Vortex analysis and various anomaly detection methods are proposed to capture flow instability that may lead to stall. Comparative visualization techniques are then employed to reveal and contrast temporal patterns from the detection results. Positive expert feedback shows the effectiveness and potential of the proposed methods for stall analysis in large-scale flow simulations.
Han-Wei Shen (Advisor)
Rephael Wenger (Committee Member)
Jen-Ping Chen (Committee Member)
190 p.
Computer Science
Large Data Analysis, Visualization, Data Summarization, Time-varying Flow Field, Out-of-core, Uncertainty Modeling, Pathline, Rotating Stall

Recommended Citations

Citations

  • Chen, C.-M. (2016). Data Summarization for Large Time-varying Flow Visualization and Analysis [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1469141137

    APA Style (7th edition)

  • Chen, Chun-Ming. Data Summarization for Large Time-varying Flow Visualization and Analysis. 2016. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1469141137.

    MLA Style (8th edition)

  • Chen, Chun-Ming. "Data Summarization for Large Time-varying Flow Visualization and Analysis." Doctoral dissertation, Ohio State University, 2016. http://rave.ohiolink.edu/etdc/view?acc_num=osu1469141137

    Chicago Manual of Style (17th edition)

osu1469141137
530
© 2016, some rights reserved.Creative Commons License Data Summarization for Large Time-varying Flow Visualization and Analysis by Chun-Ming Chen is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by The Ohio State University and OhioLINK.