Parallel Visual 3 (pV3) is an interactive computer program that provides selected displays of data generated in numerical simulations - especially simulations that involve parallel processing. pV3 was originally designed to aid in the visualization of numerical results from computational fluid dynamics (CFD) calculations on unstructured as well as structured computational meshes, but can also be used to display data from other calculations and to display intermediate computational results for diagnosis of simulation software.
Parallel processing is used to reduce significantly (relative to serial processing) the time needed to perform complex numerical simulations of engines for purposes of design or analysis. Parallel processing is becoming common in industries (aerospace, automotive, financial, and oil-exploration) that rely heavily on computational simulations. As a result, there is a need for new software tools that designers and analysts can use to interact with computational simulations while the computations are in progress. pV3 is an innovative visualization-aiding software tool that satisfies part of that need. pV3 enables the viewing, steering, and understanding of results of a simulation while the results are being generated in a distributed parallel computing environment.
pV3 was developed to support work in the emerging field of parallel application programs, as part of the participation of Glenn Research Center in the High Performance Computing and Communication (HPCC) project. HPCC began to develop large-scale simulations that exceeded the computing capacity of any current single shared-memory computer. The ability of pV3 to make visible all the data generated by distributed computers made it possible for HPCC and derivative projects at Glenn Research Center and in industry to perform simulations that they were unable to perform before.
The figure presents an example of an image of the interior of a turbofan engine that contains 5,000 airfoils, generated by pV3. Optionally, a display containing images like this one could comprise still or moving pictures. Such a display could not be generated by any other currently available software running on any currently available distributed computing hardware.
This work was done by R. Haimes of Massachusetts Institute of Technology and Gregory J. Follen of the Computing and Interdisciplinary Systems Office at Glenn Research Center. LEW-16712