A Framework for Global Scope Interactive Visual Analysis of Large Unsteady 3D Flow Data

1
A Framework for Global Scope Interactive Visual
Analysis of Large Unsteady 3D Flow Data

• Martin Ilcík
• prof. Helwig Hauser

Comenius University, Slovakia
2
Assignment

• Large simulation data
• Integral lines

3
Assignment

• Large simulation data
• Integral lines
• Interactive Flow analysis
• SimVis Doleisch 03

4
Assignment

• Large simulation data
• Integral lines
• Interactive Flow analysis
• SimVis Doleisch 03

5
Contribution

• Semi-global flow analysis
• interactive
• layered approach
• attributes structures

6
Contribution

• Semi-global flow analysis
• interactive
• layered approach
• attributes structures
• Flow feature specification
• uniform linked environment
• flow driven analysis

7
Outline

• Flow structures visualization
• Layers for visual analysis
• Framework pipeline
• Example
• Conclusions

8
Input data

• Up to 2.107 cells
• Multi-variate
• Unstructured grids
• Polyhedral cells
• Time-dependent

Dataset by Arsenal Research GmbH
9
Integral lines

• Vector field

10
Integral lines

• Vector field Streamlines

11
Related work

• Dense geometric flow visualization Park 05
• Path line attributes Shi 07

12
SimVis system

• Selective visualization
• multiple coordinated views
• focus and context
• smooth brushing

13
SimVis system

• Selective visualization
• multiple coordinated views
• focus and context
• smooth brushing
• Feature specification Doleisch 03
• derived data calculations
• feature definition language (FDL)?
• degree of interest (DOI)?

14
Flow analysis layers
15
Direct Data Analysis Layer

• Attribute values
• No geometrical information
• Brushing possible
• interval selections
• multivariate analysis

16
Local Flow Analysis Layer

• Direct neighbors
• Connectivity
• Differentials
• Eulerian point of view
• particles at a position

17
Global Flow Analysis Layer

• Lines
• Integration
• Long-term flow behavior
• Lagrangian point of view
• positions of a particle

18
Layers linking

• Bidirectional relations
• layer entities linked to cells
• and vice versa
• Degree of interest (DOI)?
• 0..1 value for each entity
• specified by brushing
• layer DOIs combined at cells

19
Layers linking
20
Framework Pipeline

• Preprocessing
• data generation
• offline

21
Framework Pipeline

• Preprocessing
• data generation
• offline

Method Interpolation Segs/s
Euler Nearest 50000
RK2 Linear 20000
22
Framework Pipeline

• Preprocessing
• data generation
• offline
• Visual analysis
• interactive
• limited by SimVis

Method Interpolation Segs/s
Euler Nearest 50000
RK2 Linear 20000
23
Framework Pipeline

• One line per cell
• Seeds in cell centers
• Regions of interest

24
Framework Pipeline

• Line type
• stream/path lines
• Accurate vs. fast
• interpolation
• numerical schema
• adaptive step size
• Direction
• Termination

25
Framework Pipeline

• Attributes creation
• segment measures
• cell measures
• Flow analysis layers
• linking
• mapping

26
Framework Pipeline

• Feature specification
• DOI, FDL
• Doleisch 03
• Coordinated multiple views
• smooth brushing
• Doleisch 02

27
Framework Pipeline

• Color mapping
• line attributes
• original data
• Orientation
• lighting Mallo '05
• halos
• focus context
• hybrid raycasting Muigg '07

28
Cooling jacket case study (1)
29
Cooling jacket case study (2)
30
Cooling jacket case study (3)
31
Summary Conclusions

• Global flow analysis
• precomputation necessary
• layered approach
• attributes structures
• interactivity

32
Summary Conclusions

• Global flow analysis
• precomputation necessary
• layered approach
• attributes structures
• interactivity
• Flow feature specification
• uniform linked environment
• flow driven analysis
• global features stronger than local

33
Future work

• Robust global feature specification
• many lines
• smart processing

34
Future work

• Robust global feature specification
• many lines
• smart processing
• GPU numerical integration
• limited memory
• unstructured grids

35
Thank you for your attention!