Fluid Flow Summarization Using Dynamic Multi-Vector Feature Spaces

Renato José Policani Borseti¹; Leandro Tavares da Silva¹; Gilson Antonio Giraldi¹

¹ National Laboratory for Scientific Computing

doi:10.20906/CPS/CILAMCE2017-0353

Resumo

Recent works have applied summarization concepts for fluid flow analysis in computational fluid dynamics (CFD) simulations to yield a synthetic and useful visual abstraction of the flow evolution. In one of these works, the pipeline firstly performs a coarse temporal stream flow segmentation that is automatically improved by k-means to complete the visual summary. The original technique considers, as the input data, only the velocity field and the particles configuration obtained by smoothed particle hydrodynamics (SPH) simulations. In this work, we demonstrate a relationship between the vorticity field and the pressure gradient. The obtained result points toward the necessity of combining the vorticity and pressure gradient as well as the velocity and particles configuration fields to compose the feature space to search for the fundamental segments of the fluid evolution. Besides, we incorporate an interval tree to improve computation of the coarse flow segmentation. We demonstrate the methodology using a 2D SPH simulation of the N-roll mill apparatus where $N=6$ symmetrically placed rollers, that rotate at constant angular velocities, are surrounded by a fluid. We show that the clusters generated by our algorithm captures a compact but detailed picture of important segments of the fluid.

Palavras-chave: Computational Fluid Dynamics; Summarization; Smoothed Particle Hydrodynamics; Fluid Visualization; N-Roll Mill Flow; Dynamic Vectors