Escolar Documentos
Profissional Documentos
Cultura Documentos
Expertise
Core CFD and CEM Algorithm development Hypersonics, internal and external flows Grid generation: IITZeus Particle methods Company: Zeus Numerix Systems and Engineering Visualization
People
A. Chatterjee (Gas dynamics, CEM, MHD) J. C. Mandal (CFD algorithms) A. G. Marathe (Internal flow, combustion) Prabhu Ramachandran (Particle methods, visualization) G. R. Shevare (Grid Generation, Industrial CFD, founder member Zeus Numerix) Krish Sinha (Hypersonics, turbulence) K. Sudhakar (MDO, design optimization)
Computational Electromagnetic solver Profs. Chatterjee and Shevare Predict the EM scattering and RCS of arbitrary, complex 3D geometries Finite volume time domain technique (FVTD) Resolving material properties using higher order schemes Grid: IIT Zeus - structured, multi-block, volume grid along with surface modeling CPU time: ???
Surface grid
Surface Current Distribution on Straight Cylindrical Cavity with Hub, Blades and Plate terminations
New algorithm: Solution dependent weighted least square (SDWLS) Higher order reconstruction for FVM on unstructured grids A new implicit method for KVFS New formulation of energy relaxation methods for Hypersonic flows Unsteady flow on dynamic meshes Multi-grid convergence acceleration methods Pseudo-compressibility methods for incompressible flow with heat transfer
Second order accurate solution dependent weighted least square reconstruction (SDWLS) No limiters required Theory unifies existing limiters and allows for generation of new ones Comparison of results for shock reflection problem (right)
Easy to extend a solver using perfect gas to compute real gas effects efficiently for inviscid/viscous equilibrium and non-equilibrium cases
ALE formulation, gometric conservation laws Upto 650 times faster than conventional explicit schemes
Grid generation
IIT Zeus Arbitrary 3D configurations Supports reading CAD surface data Fully functional GUI Surface modeling Structured, multi-block/Unstructured/Hybrid Visualization of solution on geometry
CFD Expert
Advanced fluid flow solver in 3D Developed and marketed by Zeus Numerix Full-featured GUI Built in pre and post processing Support for input from CAD data Structured multi-block/Unstructured/Hybrid Competes with the likes of Fluent, CFX etc.
CFDExpert Examples
LES of an explosion
CFDExpert ...
Pathlines for flow past a pickup truck (right) and a jeep (bottom)
CFDExpert ...
In house code Non-equilibrium thermo chemistry Turbulence models for high-Re Large scale parallel simulations 30000 CPU hours for 9 million points
Apollo shaped vehicle Flown in 1965 Mach 16, U = 5 km/s 3D unsteady wake
Vortex particles used to simulate fluid flow (Prabhu Ramachandran) Navier-Stokes fluid, incompressible, 2D Grid free, self-adaptive, unsteady flows Involves complex algorithms
Adaptive Fast Multipole methods [O(N2) O(N)] Accelerated higher order panel method Particle tracking with complex geometry
Written in C++, wrapped to Python 10 CPU commodity Linux cluster (IITM) for high-res, desktop for low-res
High resolution, initial transients for flow past impulsively started cylinder at Re=9500 Red dots: clockwise; Blue dots: anti-clockwise 200000 particles (each), 8 hours on 8 CPU Linux (P4) cluster at IITM
Flow past a complex shape 45000 particles, 5 hours on PIII (450 Mhz) Flow inside a mixer 50000 particles, 4 hours P4 (2.8 Ghz)
Visualization: MayaVi
Open Source, BSD license Cross platform: *nix/Win32/Mac Developed by Prabhu Ramachandran (partially funded by Enthought Inc.) Written in Python, uses VTK for graphics SourceForge project: http://mayavi.sf.net Supports structured and unstructured data Visualizes scalar, vector and tensor fields Support for volume rendering Interactively scriptable from Python Used world over
MayaVi
Incorporated as Pvt. Ltd. in July-Aug. 2004 Incubated in IITB's Society for Innovation and Entrepreneurship (SINE) CFDTutor : GUI tool to learn CFD CFDExpert: End-to-end tool to solve complex CFD problems CEMExpert: Tool for CEM problems Backed by consultancy and expertise of IIT faculty Competing with Fluent, CFX, etc. The only entirely Indian CFD company Alliances with CDAC and TCS
Computational resources
Galaxy.aero: 64 CPU Intel Xeon (3.2 Ghz, 2GB RAM per node) Cluster.aero: 18 CPU Intel P4 (1.6 Ghz, 1GB RAM per node) Brahma: 16 CPUs, Compaq HPC Alpha 160 (833 Mhz) Altix1: 7 CPU SGI Altix 350 Altix2 : 14 CPU SGI Altix 350 Param Padma (C-DAC, Bangalore)
Software resources
In-house tools/codes
IIT Zeus CFD Expert (Zeus Numerix ZNPL) CEM Expert (ZNPL) CFD Tutor (ZNPL) CFD/CEM/particle method codes MayaVi