Concurrent CFD Explained (Part II)
As promised, here’s the second part of the explanation of why Concurrent CFD is different, or to use a cliché, a paradigm shift in CFD.
The animation below looks more closely at why concurrent CFD is different to conventional and upfront CFD. By expanding the CFD process we can see it involves quite a number of steps. In going through the animation you will see that both conventional and upfront CFD involve transferring geometry from the CAD system and cleaning it up so it’s suitable for analysis. The process has to be repeated as design (geometry) changes are made in the analysis suite and taken back into the CAD system to keep the two in sync, so it’s inherently messy.
Typically this approach will require fluid spaces to be watertight for the analysis. In CAD terms this is referred to as ‘healing’ the geometry to make it ‘manifold’, whereas analysts often refer to it as ‘cleaning the CAD model’. This is a generic requirement for CFD analysis, so it appears in all three approaches.
You’ll see in the animation that there is a “Create Cavity” step, which requires some explanation. Most conventional CFD meshing tools work by meshing a solid object just like FEA packages do, so they require a solid object to mesh. For a CFD simulation the solid object is the flow space, which has to be created as a dummy part within the CAD system by Boolean subtraction of the entire model from an encapsulating solid. This is usually done in the CAD system and it’s this inverted flow space that’s transferred to the CFD system for meshing. Historically, conventional CFD was not designed to handle ‘conjugate’ heat transfer problems, where conduction within and radiation between solids is handled together with convection in the fluid.
Upfront CFD tools incorporating a solid modeller also require the geometry to be prepared manually for the analysis, but will make the geometry import easier and may remove the need to create the cavity manually. Meshing however remains the same.
By comparison, Concurrent CFD works rather differently. The geometry used for the analysis is native to the CAD system. This means that there is no geometry transfer step because the designer never has to leave the CAD system. As the CFD tool is embedded and has full access to all the information about the CAD model so what’s solid and what’s fluid is known, thus solid, fluid and solid-solid and solid-fluid interfaces can all be handled appropriately. Conjugate problems are handled implicitly, and the meshing is fully automated, with the user having full control of the meshing process.
Concurrent CFD therefore eliminates the “transfer geometry”, “create cavity” steps, and effectively meshes in one step. Meshing still takes place, but only takes minutes rather than hours of iterating back and forth. For example, one case I heard about recently took over a week using conventional CFD but only 10 minutes in FloEFD – a very happy customer!
Concurrent CFD provides one final benefit that’s not shown in the animation. As mechanical designers undertake their own analyses they quickly learn how to build analysis-friendly geometry within the CAD tool, eliminating the “clean geometry” step, so the time savings can be even greater than those indicated – sweet. Enjoy the animation
Dr. J, Hampton Court
- Lies, Damned Lies, and “CFD Comparison Charts” – Part IV
- Lies, Damned Lies, and “CFD Comparison Charts” – Part III
- Lies, Damned Lies, and “CFD Comparison Charts” – Part II
- Lies, Damned Lies, and “CFD Comparison Charts” – Part I
- Mechanical Analysis Products Now in Mentor’s Higher Education Program
- Sony Vaio laptop in mass ‘recall’
- Nearly Back to Business as Usual
- The Debate about Liquid Cooled Data Centers
- MicReD Technology Wins Highest Technical Honor
- Sticking Plaster and Light beats Skin Cancer
- August 2010 (4)
- July 2010 (2)
- June 2010 (1)
- April 2010 (2)
- March 2010 (4)
- February 2010 (5)
- January 2010 (6)
- December 2009 (4)
- November 2009 (5)
- October 2009 (4)
- September 2009 (3)
- August 2009 (4)
- July 2009 (6)
- June 2009 (3)
- May 2009 (2)