Http://cs.swan.ac.uk/ The Search for Swirl and Tumble Motion Robert S. Laramee Department of Computer Science Swansea University Wales, UK.

Post on 29-Jan-2016

218 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

Transcript

http://cs.swan.ac.uk/

The Search for Swirl and Tumble Motion

Robert S. LarameeDepartment of Computer Science

Swansea University

Wales, UK

2 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

Outline Problem description: swirl and tumble motion Flow visualization classification:

direct geometric texture-based feature-based

The great search 2D 2.5D (surfaces) 3D steady and unsteady flow

Summary and conclusions

3 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

Swirl and Tumble Motion swirl motion:

characterized by motion about cylinder-aligned axis

more stable (easier)

tumble motion: characterized by motion about axis orthogonal to cylinder

unstable, more difficult

4 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

Swirl and Tumble MotionAchieving ideal patterns of motion leads to optimal

mixing (of air and fuel) conditions e.g., higher exhaust/gas ratio (EGR) decrease in fuel consumption lower emissions

1. Can visualization provide insight into or verify characteristic shape/behavior of flow?

2. What tools help to visualize swirl/tumble motion?3. Where (in the combustion chamber) are ideal

ideal flow pattern not being realized?

5 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

Flow Visualization Classification1. direct: overview of vector

field, minimal computation, e.g. glyphs, color map

2. texture-based: complete coverage, more computation time, implementation time, e.g., Spot Noise, LIC

3. geometric: compute a discrete object whose geometry reflects flow characteristics, e.g. streamlines

4. feature-based: interesting subsets of the flow are first extracted

6 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

The Great SearchGeometric Texture-

basedFeature-based

2D

2.5D

3D & hybrid

Both steady and unsteady flow!

7 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2D Slices, Geometric Flow Visualization

3D streamlets for visualization of swirl motion:

Coverage is incomplete

Seeding an issue

Resampling used

swirl motion good, although

off-center

8 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2D Slices, Geometric Flow Visualization

Streamlines and tumble motion

Implemented via dye injection:

Coverage incomplete

Some flow features highlighted

Applicable to 3D Downstream

direction shown

non-ideal tumble motion, saddle

point

9 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2D Slices, Texture-Based Flow VisualizationTexture-based flow

visualization and swirl motion:

Image-Based Flow Visualization (IBFV) [Van Wijk 02]

Complete coverage Flow features shown Hard to apply in 3D downstream direction

shown

swirl behaviour evident

10 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2D Slices, Texture-Based Flow Visualization

Texture-based flow visualization and tumble motion:

Complete coverage Flow features shown Difficult to apply to 3D Downstream direction

shown

tumble behavior non-ideal

11 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2.5D, Geometric Flow VisualizationDye injection on

surface (tumble motion):

Incomplete coverage Flow features depend

on seeding Applies to 3D Downstream direction

shown

tumble motion: separatrix

highlighted

12 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2.5D, Geometric Flow VisualizationTimelines on surface: Incomplete coverage Flow features depend

on seeding Applicable to 3D Downstream direction

shown in animation

tumble motion: divergent and

convergent areas of the flow highlighted

13 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2.5D, Texture-Based Flow Visualization

Texture-based approach at surface:

Image Space Advection (ISA) [Laramee et al. 04]

Complete coverage Flow features visible Difficult to apply in 3D Downstream direction

shown

swirl motion: generally good at surface (w/

exceptions)

14 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2.5D, Texture-Based Flow Visualization

Texture-based approach on surface:

ISA Complete coverage Flow features visible Difficult to apply in 3D Downstream direction

shown

tumble motion: rotation off-center?

saddle point shown

15 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D & Hybrid, Geometric Flow Visualization

3D streamlines Incomplete coverage Flow features

depend on seeding Appropriate for 3D Downstream

direction sometimes shown

swirl motion: (generally)

prevalent inside combustion

chamber

16 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D & Hybrid, Geometric and Texture-Based Flow Visualization

Geometric and texture-based flow visualization on isosurfaces

Incomplete coverage Flow features depend on

placement Appropriate for 3D Downstream direction

shown

swirl motion: prevalent inside combustion

chamber, non-ideal motion also prevalant

17 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D & Hybrid, Geometric and Texture-Based Flow Visualization

Geometric and texture-based flow visualization on isosurfaces

Incomplete coverage Features shown depend on

placement Applicable to 3D Downstream direction

shown

tumble motion: axis of rotation is off center,

cross between two ideals

18 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2.5D Feature-Based Flow Visualization

2.5D Feature-Based Swirl Visualization

Topological skeleton of flow is extracted explicitly

Features shown automatically

Applicable to 3D Complete coverage Downstream direction not

shown

swirl motion: very complex

near top of chamber

red = saddle

blue = sink

green = source

19 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D, Unsteady Geometric Flow Visualization

3D, Unsteady Geometric Tumble Visualization

Pathlets seeded at inlet Features depend on

seeding Applicable to 3D Incomplete coverage Downstream direction

shown

tumble motion: difficult to

observe

20 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D, Feature-Based Flow Visualization

3D, Feature-Based Swirl Visualization

Boundary topology + 3D vortex core lines

Features extracted automatically

Applicable to 3D Incomplete coverage Downstream direction not

shown

Swirl motion: candidate

swirl axis off-center

red = saddle

blue = sink

green = source

21 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D, Unsteady Feature-Based Flow Visualization

3D, Unsteady Feature-Based Tumble Visualization

Boundary topology extracted + 3D vortex core lines

Features extracted automatically Applicable to 3D Incomplete coverage Downstream direction not shown

tumble motion: very transient

red = saddle

blue = sink

green = source

22 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D, Feature-Based Flow Visualization

3D, Feature-Based Swirl Visualization

Boundary topology + 3D vortex core regions-lamda 2 transfer function

Features extracted automatically, resampling necessary

Applicable to 3D Incomplete coverage Downstream direction not shown

Swirl motion: candidate

swirl axis off-center

red = saddle

blue = sink

green = source

23 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D, Unsteady Feature-Based Flow Visualization

3D, Unsteady Feature-Based Tumble Visualization

Boundary topology extracted + 3D vortex core regions –lambda 2 transfer function

Features extracted automatically, resampling necessary

Applicable to 3D Incomplete coverage Downstream direction not shown

tumble motion: complex!

candidate axis is revealed

red = saddle

blue = sink

green = source

24 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D, Unsteady Feature-Based Flow Visualization

Tumble Visualization-more difficult to visualize

Volume rendering of 3D vortex core regions –lambda 2 + helicity

Features extracted automatically, resampling necessary

Applicable to 3D Incomplete coverage Downstream direction not shown

tumble motion: candidate axis is

more clearly revealed

25 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D, Unsteady Feature-Based Flow Visualization

Tumble Motion and Hybrid Visualizations

Volume rendering of 3D vortex core regions –lambda 2 in combination with particles

Features extracted automatically, resampling necessary

Applicable to 3D Incomplete coverage Downstream direction shown

tumble motion: very transient, most effective up to

this point

26 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

3D, Unsteady Feature-Based Flow Visualization

Tumble Motion and Hybrid Visualizations

3D vortex core lines in combination with particles

Features extracted automatically Applicable to 3D Incomplete coverage Downstream direction shown

tumble motion: very transient, most effective up to

this point

27 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

2.5D, Feature-Based Flow Visualization

Tumble Motion and Periodic Orbit Extraction

Automatic periodic orbit extraction

Applicable to 3D Incomplete coverage Downstream direction shown (in

animation)

tumble motion: extracted directly,

questions with interpretation

28 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

Summary and Conclusions We have presented the most

comprehensive search for swirl and tumble motion ever…

The search continues! Future feature-based flow visualization

techniques need to be able to extract user-defined features directly, in 3D, and be applicable to unsteady flow

29 The Search for Swirl and Tumble Motion

Robert S. Larameer.s.laramee@swansea.ac.uk

AcknowledgementsThanks to the following people:

Guoning ChenChristoph GarthHans HagenHelwig HauserKonstantin MischaikowJuergen SchneiderXavier TricocheDaniel WeiskopfEugene Zhang

Thank you for your attention!Questions?

top related