Another Fine Mesh

Now might be a time to touch all your header files and kickoff a “make” to rebuild your entire project because this is a LONG post, the result of letting things build-up for weeks.

Make FUN3D Run Faster

• This may be the newsiest news item in the CFD world that I’ve seen in a while, meaning that I’ve never seen this done before. NASA’s FUN3D CFD code is the subject of two related, prize-offering challenges: The High Performance Fast Computing Challenge and the NASA High Performance Fast Computing Challenge Architecture. [Good luck remembering which is which.]
  • Both challenges are shooting for multiple order-of-magnitude run-time improvements in FUN3D on NASA SGI ICE cluster.
  • The HPFCC is an “ideation” challenge involving solver algorithm stuff. Top prize is $10,000.
  • The NASA HPFCC Architecture involves optimizing modules within FUN3D. Top prize is $15,000.
  • Important things to know: U.S. citizens only, Fortran.
• I have not reached out to my FUN3D friends at NASA, but some online discussions have questioned whether it’s even possible to eek any more performance out of a mature code like FUN3D, whether there are any people outside NASA who know enough about Fortran and CFD to even attempt the challenge, and how the NASA folks will feel if someone comes back with a 100x improvement.

Events

Screen capture from a video demonstration of XFlow CFD 2017’s new capabilities for heat transfer between air, oil, and solid. See link below.

• 19 people generated meshes for the same aircraft using the same guidelines. More importantly, they’re all going to share their meshes and [IMO more importantly] describe the resources (labor, computer) required to generate them. Want to be there to hear the “compare and contrast?” Then register for and attend the 1st AIAA Geometry and Mesh Generation Workshop, 3-4 June in Denver prior to AIAA Aviation.
• Just a reminder that you’ve still got a couple of weeks to submit your paper for this year’s International Meshing Roundtable. Papers are due 30 May and Research Notes are due 08 August.
• The ASSESS Congress 2017 will be held in Potomac, MD on 1-3 November. This invitation-only event will delve into industry and product strategies that can vastly broaden the use of engineering simulation.
• The 12th OpenFOAM Workshop is coming up later this summer at the University of Exeter in July.
• This news would have been more helpful before the event rather than after, but at the GPU Technology Conference you could’ve brought your own data to explore with a GPU-powered version of ParaView.
• CIMdata announced that Procter & Gamble’s director of modeling and simulation will deliver the keynote address at their June Product and Manufacturing Innovation Workshop.

Software

• OpenSBLI, written in Python, automatically generates C code from PDEs for numerical solution. As the name implies, the target application is Shock wave Boundary Layer Interaction using the compressible Navier-Stokes equations.
• My visualization friends will have to help me grok exactly how this would be applicable to CFD but TTK is open-source, “topology toolkit” software for “topological data analysis in scientific visualization.” See image below.
• CADfix v11 SP1 was released by ITI with updated native CAD interfaces and more.
• Tecplot released big-data, performance, and other improvements in Tecplot 360 2017 R2.
• XFlow CFD 2017 includes new capabilities for heat transfer and supersonic flows. See image above.

TTK, the topology toolkit, in action for an “uncertain starting vortex.” Image from TTK. See link above. [I applaud the use of ASCII art in their text output.]

Computers & Computing

• Hard drives could be getting a lot smaller if holmium storage technology matures in which a single atom is used to store a single bit. They’ve got a way to go, as their current implementation stores a total of exactly two bits.
• Because I ran CFD codes on a Cray-1 back in the day, Chris Fenton’s home-built, one-tenth scale, binary compatible Cray-1a makes me both happy and curious how something like this is even possible. See image below. [You do not want to know some of the things that happened to Cray supercomputers at 3am during 3rd shift, classified processing.]
• ESI and Huawei have partnered on a CAE public cloud solution.
• Also in the cloud, Envenio is now offering Pointwise via the cloud to their customers.
• ANSYS wrote about their potential to use 1 million cores.
• Agree or disagree? The coming cloud computing tidal wave.

A homemade, functional Cray-1a. Image from christfenton.com. See link above.

Applications & More

• Cambashi partnered with intrinSIM on the Cambashi CAE Observatory that shows 470 companies are contributing to a CAE simulation tools worldwide market of $5.4 billion. [Did you think either of those numbers was that large?]
• Learn about challenges and best practices for simulation preprocessing in a recording of a webinar from late April: Solving the Bottlenecks of Simulation: Preprocessing.
• Here’s a high-level (but detailed) overview of CFD in the wind power industry.
• CHAM seeks a CAD and Software Development Engineer to work on their CFD plugin-in for Rhino, RhinoCFD.
• In other Rhino news, Mesh2Surface 4.3 (reverse engineering plugin) was released.
• See how to use midas NFX CFD to simulate the flow in a manifold.
• Read how CFD is being used to simulate ship propellers including cavitation. See image below.
• Tech Clarity is offering a free [and awkwardly titled] e-book on Adopting a New 3D CAD. Registration required. [I just downloaded it and will read it over the weekend because I’m wondering if the ideas could be genericized as “Adopting a New 3D <insert software type here>”.]

CFD simulation of a ship propeller. Image from The Maritime Executive. See link above.

Science to Art

Just so you don’t think I see only meshes everywhere, here’s the work of Jonathan Feldschuh who paints scientific simulation results. He writes about the digital approximations inherent in simulation results (i.e. mesh resolution, pixellated images) and paint’s fluid ability to interpolate and create an analog version of the same results. In other words, using a real fluid (paint) to represent a simulation of a real fluid.

Jonathan Feldschuh, Simulation of a SN1a blast interacting with a main sequence star (left), SN1a.1.01, acrylic on canvas, 2004 (right).

Bonus: Thanks to valued reader Joe, I can share with you the work of Frederico Diaz who used a RealFlow fluid simulation as the basis for his installation at Mass MOCA of Geometric Death Frequency-141.

Frederico Diaz, Geometric Death Frequency-141. See link above.