Scalable methods for visualizing flow in a pellet filled reactor.

Jaipreet Singh
M.Tech. Thesis


Smoothed-particle Hydrodynamics (SPH) simulations generate large amounts of flow field data. Extracting knowledge from these volumes of data and visualizing the huge data are challenging problems as the simulation produces a collection of particles together with multiple physical quantities like density, velocity, pressure and temperature at each particle location. For simulating flows of fluid in a reactor filled with pellets, our aim is to visualize 100 million particles. As a first step, we develop a scalable parallel algorithm to convert the simulation output into a grid representation that is amenable to fast visualization. The parallel algorithm is implemented on a manycore (GPU) architecture. In the next phase, we develop fast visualizing techniques that helps in interactive visualization of the generated 3D grid.