Talking CFD

We could talk about cloud CFD until the cows come home (& we did) so we've split this topic in two. In this first part, we start by trying to answer the question - what do we even mean by "cloud?" - before going on to elaborate on some of the differences between doing CFD in the cloud and in-house.

Simon picks me up on why I think "bursting into the cloud" is nonsense. And we discuss whether larger companies are really ready to go all in on cloud CFD?

This week's question is - "if people understand the benefits of using the cloud for CFD, then why aren't more people actually doing it?"

Checkout....

• The Beginner's Guide To OpenFOAM on AWS @ http://learn.cfdengine.com

• AWS Snowmobile - https://youtu.be/8vQmTZTq7nw?t=45s

This second instalment of the Social can only be described as a fanboy-episode. Simon & I coo over ParaView whilst discussing post-processing in general. We touch on what we need from a post-pro package, how we do our post-pro & why we're so fond of ParaView.

I guess the question of the episode is - "when an open-source package is this good, why ever would you use anything else?"

Find us on the socials or in the Facebook Group (www.talkingcfd.com/fb) and let us know your take.

Check out – https://www.paraview.org/

Not your usual Talking CFD episodes & definitely not an interview. Rather it's a series of conversations between two friends who’ve both been CFD-ing for a while & take an interest in what’s been, what’s what & what’s to come in CFD.

Topics vary from show to show, but over the course of a few episodes we’re likely to give our take on (amongst other things):

• The current state of CFD in the cloud

• Whether our vision of CFD in 2030 is anything like NASA’s

• Just how much we really like ParaView

Hopefully you’ll dig it - let us know on the socials

Shownotes

• Connect with Simon on Twitter and/or LinkedIn

- https://twitter.com/SiHubbard

- https://www.linkedin.com/in/simon-hubbard-48b96310/

• Connect with Robin on LinkedIn - https://www.linkedin.com/in/robinknowles/

• Join the Talking CFD family on Facebook – www.talkingcfd.com/fb

It’s always a pleasure to talk to a guest that has been at the top of our industry long enough to help shape how it looks today. Professor Charles Hirsch founded NUMECA 25 years ago, and has grown it into one of the pillars of commercial CFD, particularly in the maritime sector.

It’s difficult to distil 25 years into a 30-minute interview, so we don’t even try. Instead, Professor Hirsch gives us the benefit of his experience on:

• how they’ve been “appifying” CFD since before it was trendy

• working hand-in-hand with academia for R&D

• how they tackle the eternal CFD balancing act — accuracy vs. reliability vs. robustness

• why Lattice Boltzmann is so hot right now

Shownotes

https://www.numeca.com/ – All about NUMECA

https://youtu.be/7hVlwSpYe-E – NUMECA and the America’s Cup

In this episode I’m joined by Sebastian Desand & Johan Hoffman, two of the founding team behind Ingrid Cloud. If you haven’t yet crossed paths with Ingrid Cloud, it’s an innovative new cloud CFD platform that’s driving towards CFD without manual intervention.

We talk briefly about the technology & the platform (it was very new when we recorded this), and also discuss:

• how Ingrid Cloud fits within the wider landscape of Industry 4.0

• what CFD with no manual input might look like

• whether our ideas of start-of-the-art CFD are stuck in a cul-de-sac & what the new state-of-the-art might be

Sometimes things move quickly in our industry. As such, I’d recommend the interested listener to check out Ingrid Cloud as it’s undoubtedly changed a lot since this interview.

Shownotes

https://www.ingridcloud.com/ - for the latest on the offering & to try out the demo

Connect with the guys on LinkedIn:

• Sebastian Desand - https://www.linkedin.com/in/sebastiandesand/

• Johan Hoffman - https://www.linkedin.com/in/johan-hoffman-4219041/

It's always good for me to speak to other CFD consultants. It's an opportunity to compare notes & to make sure that I'm not missing out on a "magic bullet" that no one told me about.

Today I'm talking to Stewart Bible of Resolved Analytics about how he built a CFD consultancy, sold it & then built another one. We chat about:

• the acquisition process

• bridging the credibility gap

• the importance of investing in your own IP

• an interesting take on the relative merits of commercial vs. open-source for CFD consultants

As always, you can't help but learn when people have been there and done it - take a listen.

Shownotes

• Find out more about Resolved Analytics – https://www.resolvedanalytics.com/cfd-consulting/

• David Fields' Consulting book – http://amzn.eu/hmCZrW1

* Connect with Stewart on LinkedIn – https://www.linkedin.com/in/stewartbible/ & mention the podcast

I enjoy all the interviews, but I particularly enjoyed this one — part-business, part-history lesson, in a good way. I’m joined by Eugene de Villiers to talk about the company he co-founded – Engys.

Alongside discussing Engys, it’s products and services, we also dig into the wider world of OpenFOAM, including:

• Some of the early history, before FOAM was Open

• Whether the current “splitered” OpenFOAM ecosystem could actually be a good thing?

• Whether another open-source CFD code could hope to displace OpenFOAM?

I naturally gravitate towards OpenFOAM-related companies, it’s a self-acknowledged bias on my part. But I think you’ll get something out of this interview whether you have an interest in OpenFOAM or not - enjoy.

Shownotes

- https://engys.com/ - open-source CFD for enterprise

- https://engys.com/products/helyx - their flagship product

- https://www.linkedin.com/in/eugene-de-villiers-81833722/ - Eugene on LinkedIn

This episode is both, the second GPU-related show of the season, and also a cloud simulation episode. The company on today’s show is innovating on both those fronts with their GPU CFD code and their cloud simulation platform.

Today I’m talking with three of the key people behind the Bristol-based, CFD innovators —Zenotech— Jamil Appa (co-founder), David Standingford (co-founder) & Mike Turner (product lead). They take it in turns to fill us in on:

• The scale of the massively parallel architectures at play in HPC today;

• Whether the previous generation of CFD codes keep up or if we need to rearchitect our codes as well as our hardware?

• Do we still need CPUs or can we move over to GPUs entirely?

• How they help users navigate the ever-growing number of cloud computing providers

• And how they can even make suggestions as to when & where you might like to run your CFD jobs

As previously, please forgive any GPU 101 questions — we’re still learning here.

Shownotes

https://zenotech.com/ – Simulation Unlimited

https://zenotech.com/zcfd-zenotech-computational-fluid-dynamics/ – GPU CFD

https://zenotech.com/epic-elastic-private-interactive-cloud/ – Cloud CFD

No season of this podcast would be complete without one or two OpenFOAM-related episodes, so here we are. I’m talking with Luboš Pirkl, one of the co-founders of CFD Support, a multi-faceted CFD company based in Prague, Czech Republic.

We dive into some of the usual stuff that comes up when we talk OpenFOAM, but Luboš also gives us some extra insight into:

• the difficult early days & how close they came to giving up

• how the CFD marketplace has moved away from dedicated CFD engineers & that OpenFOAM isn’t a great fit for these new users

• the decisions behind creating a vertical CFD application and why they chose the turbomachinery market

• why communicating the value of your product or service is often more difficult than any of the technical aspects

Shownotes

• https://www.cfdsupport.com/

• https://www.cfdsupport.com/turbomachinery-cfd.html

New Episode – New Tech. In this one we're looking at the move towards GPUs for compute & talking with Envenio - one of the companies looking to take maximum advantage of this emerging hardware trend for better CFD.

I’m not super-familiar with what you can (and can’t) do with a GPU in a CFD context - so forgive me if I ask some 101-level questions. Ian McLeod (CEO) & Scott Walton (VP Business Development) kindly fill me in on:

• where the trend towards GPU compute is comping from and where it might be going

• does new architecture need new “clean-sheet” software?

• what sort of speedup can we expect from this new hardware-software combination?

• some of the marketing lessons they learned trying to build up a new brand & reach CFD users

Shownotes

www.envenio.com for more about what they do & to book a free demo

blog.envenio.com/surveyresults2 to find out more about the CFD Users Survey that Scott mentions in the episode