CFD Validation on FAITH Hill
NASA experimentalists are gathering data for flow over Fundamental Aeronautics Investigates THe Hill, aka FAITH Hill. The data for this asymmetric hill will be used for validation of CFD methods.
Survey on Engineering Software Market Trends
Tech Clarity and Cyon Research are asking for your help in identifying the top trends in the engineering software market. The results of a brief survey will be revealed at COFES, Conference on the Future of Engineering Software, to be held 12-15 April.
Last year’s results indicated the top priorities were to grow in both existing and new markets, get new partners, and create new products. In technology trends, 39% cited more integration of design and analysis software (the leading vote getter). This year’s focus will be less about the overall economy and more about the long term.
- 2011’s trends on Tech Clarity
- 2012’s survey
University of Houston Student Wins CD-adapco Contest
Leonardo Chica, a masters student at the University of Houston in mechanical engineering technology, was chosen as the winner of CD-adapco’s Academic Paper Contest. Chica’s work, Fluid Structure Interaction Analysis of Two-Phase Flow in an M-shaped Jumper, will be presented at the STAR Global Contest 2012 for which he was awarded an all expenses paid trip.
- CD-adapco’s announcement
Work Smarter, Not Harder – An Optimization Story
This post on the Frozen Fractal blog (about video game development) isn’t exactly fresh (it’s about 2 years old) but its lessons about optimizing a fluid simulation and achieving a 7.6 times speedup never get old. The section about pointer aliasing was very interesting.
- Optimization Story from Frozen Fractal
Reverse Engineering Turbine Geometry for CFD
Hydro-Quebec uses 3D scanning to obtain accurate as-built geometry data on which CFD and other simulations can be performed to measure performance and resolve operational problems. In one instance they were able to increase turbine efficiency by 1.5% by changing the blade’s trailing edge geometry when simulation revealed recirculation.
Win € 1,000 in the Palabos 2012 Challenge
FlowKit, the firm that manages the open-source CFD code Palabos, is offering a first prize award of € 1,000 for the best numerical simulation based on Palabos in their Palabos 2012 Challenge. Four prizes are being offered in the challenge which is designed to increase the public knowledge of this lattice Boltzmann-based CFD code.
News in Brief
- Jenny Chueng from the University of Loughborough won the Best Overall Performance award for work she did on Low Carbon Building Design and Modeling. This course in the University’s Department of Civil and Building Engineering involves CFD and energy modeling of structures.
- Murphy Prototypes employed CFD in the design of their car for the European Le Mans Series of races.
- More about the overset mesh capability in STAR-CCM+ v7.
- FLOW Science released FLOW-3D Cast Version 3.5 for casting process simulation.
- CD-adapco released the full agenda for the upcoming STAR Global Conference.
- CFD helps reveal propeller cavitation.
- Bellco Ventures introduced the Air Mouse, an $80 glove with sensors with which you control your 3D CAD software or mesh generator. [Note: This should’ve been included in last week’s list of new interaction devices.]
- ANSYS CEO Jim Cashman cites automotive as a rapidly growing market in an interview with Fox business.
- C&R Technology and SpaceClaim announced the first advanced FEA mesher built into SpaceClaim’s directly modeling CAD software.
- It comes as no surprise that Ferrari used CFD to increase the down force and decrease the drag on their new 73o hp, 212 mph F12berlinetta.
- ESI released a new version of their ACE+ Suite including CFD.
- Reaction Design announced the addition of soot particle size capabilities to its FORTE software for combustion CFD.
- The improvement of nasal airway ventilation by rapid maxillary expansion was detected by CFD.
- Blog post: Why visualization is important for CFD.
- CFD helped show that air curtains are 10% more energy efficient than vestibules.
- Symscape provides a sneak peak at their Caedium v4 CFD software.
Reproducible Science Requires Open Source Software
This week’s conversation starter is an editorial in Nature that takes the position that reproducibility suffers when proprietary software is used in research. Just as experimentalists provide detailed description of laboratory appartuses and procedures, source code, not descriptions of source code, should be submitted for review as part of any publication of computational results (or so they say). Of course, as anyone who writes software for a living knows, any publication of flow charts or equations is virtually worthless when it comes to reproducing a computer program. There is only one true source [pun intended]. Or as Star Wars fans say about documentation, “Use the source, Luke.”
Some interesting barriers to the publication of source code are cited. Universities and national labs are always trying to monetize their software. Researchers try to maintain a competitive advantage. And it appears that some researchers may be reluctant to release “ugly” source code for fear of embarrassment. At least one person with a sense of humor addresses this last issue with a proposal to create CRAPL, the Community Research and Academic Programming License to absolve code authors of that shame.
The argument boils down to this, if reproducibility is a core tenet of science and software can only be reproducible with source code, the code must be open source.
Your thoughts?
I’ve been on both sides of the open source software issue, having developed both closed an open source codes. Some things I’ve learned:
– in a research environment, it’s very rare that being closed source is good for your funding; being open, and making sure it can be used “hands off”, improves your chances to gain critical mindshare, exposes you to interesting problems you wouldn’t do yourself, gives you ammunition for new research you want to do, and, of course, informs you of bugs sooner
– the type of open source license you use matters; people very often confuse the LGPL license as being viral, like the GPL license, even though (IMO) LGPL is the preferred license for a researcher to use. There are many projects that use the BSD license, but I view that as a license to do just about anything.
– For releasing code, it’s critical to have a click-download process that places minimal restrictions on what the downloader must do, i.e. don’t use a long information form they need to fill in; in practice, a long form that needs to get signed by somebody is just as bad as charging a fee, and it doesn’t matter if the fee is small (just getting a check from your institution is hard enough).
– For tracking users alone, it’s best to have official releases and downloadable tarballs; exposing an anonymous-read repository is fine, but makes it impossible to track download statistics
And finally, IMO, HPC applications wouldn’t be where they are today without open-source software, and that goes way beyond just solvers (e.g. it includes MPI implementations too).
Hi Tim. Thanks for sharing your thoughts on open source software. I infer from your first bullet item that you’re in favor of making source available as a condition of publication in scholarly journals.
Your second bullet complicates that, however. U.S. government software usually comes with restrictions on redistribution that would only complicate matters. And because government researchers are likely authors…
Certainly the contributions of open source to HPC are vast. It makes for an interesting debate about the relative merits of closed versus open source. But you say that being open source gets you bug reports sooner. My experience is that people are very motivated to report bugs in something they’ve paid for.
Thanks again for contributing to the discussion.