Vortex Detection on Unstructured Grids - PowerPoint PPT Presentation

1 / 14
About This Presentation
Title:

Vortex Detection on Unstructured Grids

Description:

Vortical fluid flow applications. Airplane wing design. Naval propeller design ... Non-vortical direction spanning flow 'Swirl Vector' Computation. Two possible ' ... – PowerPoint PPT presentation

Number of Views:63
Avg rating:3.0/5.0
Slides: 15
Provided by: monikajan
Category:

less

Transcript and Presenter's Notes

Title: Vortex Detection on Unstructured Grids


1
Vortex Detection on Unstructured Grids
  • Monika Jankun-Kelly
  • Mississippi State University ERC
  • Simcenter

2
Motivation
  • Vortical fluid flow applications
  • Airplane wing design
  • Naval propeller design
  • Bronchial airflow modelling
  • Goal understand vortex behavior
  • Detect and verify
  • Characterize
  • Track

3
What is a Vortex?
  • Swirling flow
  • Velocity vectors are tangent to a helical path
    around a vortex core

4
Detection Algorithm
  • Ming Jiangs combinatorial vortex detection for
    structured grids adapted for unstructured grids
  • For each cell
  • Calculate swirl vector at cell center
  • Project velocity vectors at nodes onto swirl
    plane
  • Check for direction spanning

5
False Positives
  • Sources and sinks
  • Non-vortical direction spanning flow

6
Swirl Vector Computation
  • Two possible swirl vectors
  • Vorticity vector, curl of velocity field
  • Real eigenvector of velocity gradient tensor

vorticity
real eigenvector
7
Velocity Gradient Computation
  • Structured grids have explicit connectivity and
    permit use of simple central difference at nodes
  • Unstructured grids have implicit connectivity
    (node-node neighbors not specified)
  • Unstructured grid gradient computed at cell
    center using coordinate transform method

8
Results
  • Vortices detected
  • False positives

utv propeller dataset
p5168_rfrm propeller dataset
9
False Positive Elimination
  • Surface cells not considered (ineffective)
  • Stronger spanning requirement
  • Velocity vectors rotate in same direction
  • Cell size thresholding
  • Aggregate size thresholding (unimplemented)

10
Stronger Spanning Requirement
3 out of 4 quadrants
4 out of 4 quadrants
  • 3 out of 4 quadrants is just right in most cases
  • 4 out of 4 is too strict
  • 2 out of 4 is not strict enough

11
Same Rotation Direction
all vectors have same direction
most vectors have same direction
  • Same direction for all, too strict
  • Same direction for most, can be helpful

12
Cell Size Thresholding
50th percentile
5th percentile
  • Cells in regions of interest tend to be small
  • Effectiveness of depends on parameter
  • Unscientific hack, but makes pretty pictures

13
Aggregate Size Thresholding
  • Neighboring vortex core cells aggregate
  • Cells in aggregates of size lt N eliminated
  • Used in structured code
  • Currently unimplemented in unstructured

14
Future Work
  • Near future
  • Compare results with structured algorithm
  • Verification
  • Characterization
  • When time permits
  • Different gradient computation methods
  • Other file formats (CGNS, Cobalt)
  • Long term goal
  • tracking
Write a Comment
User Comments (0)
About PowerShow.com