Visualization and Turbulence Analysis of a Particle-laden Turbulent Jet

RESEARCH                                                 RESUME                                             PHOTOGRAPHY                                       KNOW ME

The goal of this work was to attain scientific discovery through effective visualization. The following points summarize our work:

• Implemented a binning technique to the jet particle location dataset, to introduce opacity. 
• Investigated fluid structures formed in the particle laden turbulent jet for different Stokes number using a simple OpenGL visualization technique and the existing 2 Phase Direct Volume Rendering software, currently being used to visualize sonar (acoustic) data of hydrothermal plumes.
• Studied the effects of changing resolution and changing parameters of the 2 Phase DVR software on recognising fluid structures.
• Studied the effect of the dispersed phase (particles) on the continuum phase of the turbulent jet, with focus on the mean axial velocity, velocity fluctuations and turbulent kinetic energy.

INTRODUCTION:

Most flows in day to day life are made up of just one phase , the continuum phase (the fluid). A multiphase flow consists of a continuum phase, interspersed with a discontinuous phase. Examples of muti-phase flows are solid propellant rockets, plasma spray coating, volcanic ash and hydrothermal plumes (click for more info on hydrothermal plumes). The phases can be either One-way coupled where particles do not affect the continuum phase or Two-way coupled where particles and continuum phase interact mutually affecting each other. This study was motivated  to better understand the behavior of multiphase flows so that industrial processes can be controlled accurately and the natural processes and events related to multi-phase flows can be predicted more accurately.

Fluid dynamists develop numerical simulations to understand the behavior of particles and the continuum phase in a turbulent jet. We used resultant data of a similar direct numerical simulation developed by Kottam. His work simulates a 3D particle laden turbulent jet, modeling a dilute particle laden flow with a volume fraction of the order of 10^-3 and the Reynolds number of 2400. Particles are inertial and their Stokes Number(St), which is the ratio of the particle response time to the characteristic flow time, varies from 0.01 to 10. 

VISUALIZATION:

To explore the fluid structures and dispersion of particles, longitudinal slices of the particle laden jet were used in Kottam’s thesis as shown in the figure below. Since the simulation consists of various densly populated regions in the jet, it is hard to recognise fluid structures because of particles in the forefront occluding the view. 

Longitudinal slices of the particle laden turbulent jet by Kottam

By binning the particles into voxels we introduced opacity to the dataset. This reduced occlusion by other particles and made exploration easier. Since the granularity was reduced by binning, in future we would like to explore if we lost any fluid structures during this transition from visualizing particles to visualizing voxels. The following figure shows a comparison of visualizing the turbulent jet using three different softwares.

The 2 Phase DVR and Paraview visualize the binned data and clearly show formation of fluid structures. Tecplot on the other hand directly visualizes particles and as expected occludes most of the structures. Using OpenGL we visualized the voxels by a simple method just to show how even the most primitve rendering methods can help focus on fluid structures. The following video shows fluid structures as we travel from the bottom of the jet to the top, looking down.

The following figure shows how varying the resolution can affect detection of the fluid structures. As we approach finer resolutions the structures are more and more distinct. But in order to retain this advantage of opacity we have to limit the resolution to a certain level. Otherwise as we try to visualize finer and finer voxels, the number of particles in each voxel will tend to a number which is same in all voxels in view. This will nullify the purpose of binning the particles in the first place.

Scientific discovery through visualization:

One of the scientific goals of this study was to analyse the change in structure formation as the Stokes number changes. The Stokes number is a measure of the particle inertia. When the Stokes number is smaller than 1, particle motion is tightly coupled to the fluid motion, consequently the particle dispersal is the same as the fluid dispersal.  When the Stokes number is large, that is more than 1, the particles are not influences by the fluid and pass through the flow without much deflection in their initial trajectory. The interesting case is where the Stokes number is about 1, when the particles migrate to the margins of the eddy.

In the flow for St = 0.01 particles inside the bigger vortices are found to accumulate themselves in the form of smaller circular structures. At St = 0.1 some particles still follow the small circular structures but most of the particles start to travel towards the periphery of the vortical region. For St 1 the particles are flung out to the edge of the large structures, accumulating near the edge of the vortices with maximum lateral dispersion. Smaller structures are diminished at this stokes number. At St = 10 particles are seen to be unresponsive to the fluid flow.

In future we would like to investigate the formation of spirals by measuring quantities such as Swirl strength, spiral cross-section and pitch. We would also like to study the different resolutions of turbulent jet images to corelate and investigate structures in hydrothermal plumes.