- Here’s Kenneth Wong’s article from digitalengineering247.com that I’ve been silent on for a while despite a lively conversation on LinkedIn: The Black Box Dilemma: Trust The Mesh Generated by the Software or Take a Proactive Approach? My silence is the result of reading and re-reading and thinking about the article.
- The answer to his [perhaps rhetorical] question is an unequivocal “Yes. Trust the mesh.” We must make meshing invisible to the user. Many meshing software packages are at various stages of moving in this direction.
- Understandably, there’s a bit of fear involved in taking this leap of faith. That fear can and must be assuaged through rigorous verification and validation.
- Some of the article’s commentary reflects something I heard or read [and forgot the source but it might be the book The Mythical Man-Month] about automation which goes something like this. People want things done automatically but also done exactly how they would’ve done them by hand. We need to let this go and trust the automatic results based on our rigorous V&V.
- I would further propose that we stop thinking about a mesh’s quality and instead consider its suitability for a particular purpose which includes everything from resolving the geometry model to the cell shapes and sizes and their implications for the flow solver’s numerical algorithms, and the number of cells used to resolve specific regions of the flow to a desired accuracy.
- While we certainly want to be economical, if meshing consumes 75% of a person’s labor for each CFD simulation (a commonly accepted number) how important is it that the computer (a depreciating fixed asset) has to run a little longer if we use too many points? Besides, doesn’t the cloud provide us with cost-effective access to virtually infinite computing power? Through V&V we can develop best practices for the resolution to use.
- The concept of suitability implies that one black box won’t adequately serve all users, that the automated tool must allow for a user’s best practices (arrived at through rigorous V&V). At least this will have to be done to address each particular flow solver’s idiosyncrasies. See above comment about automating what would’ve been done manually.
- I’m glad Kenneth wrote this article to shed light on automatic meshing and the conversation on LinkedIn proved how important this is to a lot of people. Automatic by itself means nothing because anyone can create an automatic mesher that produces meshes that are virtually useless. I’m fairly certain it was Fred Brooks, author of The Mythical Man-Month who wrote that the word automatic has always been a synonym for “better than we’re currently doing.” With respect to him, we need to move past that to the dictionary definition, without human intervention, while also maintaining suitability.
Speaking of meshing…
- In the on-demand webinar High-Order Mesh Generation Using Pointwise you will learn how to curve and elevate a linear mesh up to polynomial degree four (Q4) using the classic Rotor 37 test case. You’ll also see some of the software’s newest capabilities in action.
- Pointwise has an entry-level opening for an engineer on our Technical Support Team that’s perfect for new grads who love CFD.
- Pointwise is also seeking engineering interns for summer 2019 on our Technical Support, Product Development, and Applied Research teams.
- A Lattice Boltzmann CFD solver was used to simulate blood flow throughout a full-body, patient-specific, arterial system.
- Tecplot 360 was used to look at CFD simulations of stock car external aerodynamics.
- ADS has been awarded USAF funding to commercialize several technologies including a GPU-accelerated flow solver.
- Background on SimScale’s new LBM flow solver called Pacefish is covered in this article from Design News.
- Researchers at UC San Diego have released OpenLSTO for level set based structural topology optimization.
- OpenFOAM is now available on AWS and Azure.
- MSC Nastran 2019 was released.
- The HPCMP’s CREATE software has been packaged for educational use as CREATE Genesis.
- Cobalt V8.0 was released with several new features including the QCR addition for Spalart-Allmaras turbulence models.
News & Events
- The ISC High Performance Conference 2019 for HPC will be held in Frankfurt on 16-20 June 2019.
- For those of us who missed the 13th International Conference on CFD in the Minerals and Process Industries the folks at Applied CCM have shared their perspectives.
- ANSYS shared the winners of their 2019 Hall of Fame competition and from those I think my favorite is from Linkoping University [sorry about the umlauts] and their simulations of flow in the human heart.
- Democratization is cited as common trend for 2019 that underlies all the buzz-wordy trends like Digital Twin, 3-D printing, and the like.
- ProNET Update, the newsletter for the Code_aster professional network, is now online.
- Siemens PLM Software has an opening for a Software Engineer in specialized meshing.
- BAE Systems seeks a CFD senior engineer.
On the Fringe
- Technology trends affecting businesses in 2019 begins with 5G, touches briefly on quantum computing, and penultimately ends with the exodus of tech firms from Silicon Valley where the average home price is nearly $950,000. [I thought for quite a while on whether to share this article here but decided the tech focus was sufficient. Who knew I actually thought while blogging?]
- Again, kinda on the fringe of CFD content but if you’ve ever tried to format mathematics on the web you can probably guess how useful CopyPasteCharacter.com is so you can do stuff like this ⦛ ∮ ≑
A Mesh We Can All See This Summer
Let’s bookend this post with another work by someone named Kenneth. Kenneth Wong started us with meshing and Kenneth Showell brings us to a close with Besped from the collection at the Blanton Museum.
The good news is that with the U.S. Congress on Computational Mechanics, including MeshTrends and a mini-symposium on mesh adaptation, happening in Austin this summer we can all go visit the Blanton Museum and see Showell’s curvilinear structured quad grid in person.
I haven’t been able to find out much about Showell’s influences from the artist’s website so we’ll just have to enjoy his work at face value for now.