A novel method for unsteady flow field segmentation based on stochastic similarity of direction
Recent developments in fluid dynamics research have opened up the possibility for the detailed quantitative understanding of unsteady flow fields. However, the visualization techniques currently in use generally provide only qualitative insights. A method for dividing the flow field into physically...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
AIP Publishing LLC
2018-04-01
|
Series: | AIP Advances |
Online Access: | http://dx.doi.org/10.1063/1.5026152 |
id |
doaj-99ae72e497db46668d9609edcdad5d33 |
---|---|
record_format |
Article |
spelling |
doaj-99ae72e497db46668d9609edcdad5d332020-11-24T22:24:22ZengAIP Publishing LLCAIP Advances2158-32262018-04-0184045020045020-1210.1063/1.5026152070804ADVA novel method for unsteady flow field segmentation based on stochastic similarity of directionNoriyasu Omata0Susumu Shirayama1Department of Systems Innovation, The University of Tokyo, 7-3-1 Hongo, Bunkyo-Ku, Tokyo 113-8656, JapanDepartment of Systems Innovation, The University of Tokyo, 7-3-1 Hongo, Bunkyo-Ku, Tokyo 113-8656, JapanRecent developments in fluid dynamics research have opened up the possibility for the detailed quantitative understanding of unsteady flow fields. However, the visualization techniques currently in use generally provide only qualitative insights. A method for dividing the flow field into physically relevant regions of interest can help researchers quantify unsteady fluid behaviors. Most methods at present compare the trajectories of virtual Lagrangian particles. The time-invariant features of an unsteady flow are also frequently of interest, but the Lagrangian specification only reveals time-variant features. To address these challenges, we propose a novel method for the time-invariant spatial segmentation of an unsteady flow field. This segmentation method does not require Lagrangian particle tracking but instead quantitatively compares the stochastic models of the direction of the flow at each observed point. The proposed method is validated with several clustering tests for 3D flows past a sphere. Results show that the proposed method reveals the time-invariant, physically relevant structures of an unsteady flow.http://dx.doi.org/10.1063/1.5026152 |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Noriyasu Omata Susumu Shirayama |
spellingShingle |
Noriyasu Omata Susumu Shirayama A novel method for unsteady flow field segmentation based on stochastic similarity of direction AIP Advances |
author_facet |
Noriyasu Omata Susumu Shirayama |
author_sort |
Noriyasu Omata |
title |
A novel method for unsteady flow field segmentation based on stochastic similarity of direction |
title_short |
A novel method for unsteady flow field segmentation based on stochastic similarity of direction |
title_full |
A novel method for unsteady flow field segmentation based on stochastic similarity of direction |
title_fullStr |
A novel method for unsteady flow field segmentation based on stochastic similarity of direction |
title_full_unstemmed |
A novel method for unsteady flow field segmentation based on stochastic similarity of direction |
title_sort |
novel method for unsteady flow field segmentation based on stochastic similarity of direction |
publisher |
AIP Publishing LLC |
series |
AIP Advances |
issn |
2158-3226 |
publishDate |
2018-04-01 |
description |
Recent developments in fluid dynamics research have opened up the possibility for the detailed quantitative understanding of unsteady flow fields. However, the visualization techniques currently in use generally provide only qualitative insights. A method for dividing the flow field into physically relevant regions of interest can help researchers quantify unsteady fluid behaviors. Most methods at present compare the trajectories of virtual Lagrangian particles. The time-invariant features of an unsteady flow are also frequently of interest, but the Lagrangian specification only reveals time-variant features. To address these challenges, we propose a novel method for the time-invariant spatial segmentation of an unsteady flow field. This segmentation method does not require Lagrangian particle tracking but instead quantitatively compares the stochastic models of the direction of the flow at each observed point. The proposed method is validated with several clustering tests for 3D flows past a sphere. Results show that the proposed method reveals the time-invariant, physically relevant structures of an unsteady flow. |
url |
http://dx.doi.org/10.1063/1.5026152 |
work_keys_str_mv |
AT noriyasuomata anovelmethodforunsteadyflowfieldsegmentationbasedonstochasticsimilarityofdirection AT susumushirayama anovelmethodforunsteadyflowfieldsegmentationbasedonstochasticsimilarityofdirection AT noriyasuomata novelmethodforunsteadyflowfieldsegmentationbasedonstochasticsimilarityofdirection AT susumushirayama novelmethodforunsteadyflowfieldsegmentationbasedonstochasticsimilarityofdirection |
_version_ |
1725761734087540736 |