Anyone who keeps up with simulation data management understands that the increase in both the size and number of CAE simulations that engineers and designers are managing is a huge area of concern.
How huge?
Tsunami-huge.
Don’t worry, though; we’re right on top of that.
Dr. Scott Imlay, Director of Research at Tecplot, Inc. recently travelled to surfing Mecca, San Diego, California to present a paper about the need for a new approach to data management and visualization at the Engineering Design Forum 2012. The conference brought together users, researchers, and developers to explore critical issues in simulation-based design.
At issue: more and more data to manage
A quick description of the problem: During the engineering design process, engineers run a lot of simulations, including CFD calculations, finite element analysis, and much more. Not surprisingly, these simulations create enormous amounts of data. High-fidelity cases can run into the hundreds of millions of cells. Even a low-fidelity case can run into the tens of millions of cells. As engineers work through a project, they vary many of the parameters so as to better understand the behavior of their systems. Each parameter then becomes a dimension in the data.
“We have customers who are looking at as many as 93 dimensions they need to understand,” said Scott. “Among those who are doing geometric optimization, the number of dimensions skyrockets.”
A new, holistic way of dealing with simulation data of highly dimensional space
At many companies, engineering and design are conducted just the way they were in the early 90s. Back then, post-processing analysis capabilities were designed to analyze just one simulation case at a time. Though things have changed a lot since then, not all designers and engineers have kept up with those changes.
“Too many designers write their own scripts to generate extractions,” Scott observed. He added that most designers are not employed to be scripting experts. Therefore, what’s needed is an automated solution.
Well—as you may have guessed—we’ve got one of those.
A data mining approach
Engineers and designers need some mechanism for selecting data out of their data sets, then analyzing it over their entire collection. That way, they can see how that “extracted data” compares over their design space.
Tecplot Chorus allows just this. It enables engineers to couple field data and meta data for faster analysis and understanding. It also allows them to extract certain views of that data so that they can look at a large number of cases simultaneously and get a much better feel for the field data. The software also makes downstream analysis much faster, changing the post processing paradigm by pre-computing the plots.
In short: it’s an entirely new way to work.