Flowkit CEO, Jonas Latt (Part 1)

I like my CFD in the sky.If I were a betting man, I’d place some money on the Lattice Boltzmann Method (LBM) to eventually shake up the CFD world. There are a few companies focusing on LBM, but I’d also be sliding a few side bets on Flowkit having a major hand in its growth.

Flowkit is a lean, young CFD company enjoying the luxury of building new technology from scratch… without the baggage of decades of existing code/customers, and with the cloud in mind from day 1.

I met Jonas Latt, CEO of Flowkit, about a year ago while researching CAE companies with a focus on the cloud. Jonas educated me on the advantages of Open Source in the CAE space, as well.

In part 1 of this 2 part interview, we’ll dip our toes into LBM and Open Source CAE.

Links from this episode:

FlowKit CFD Solutions
Palabos Open Source CFD



  1. Hi Jeff: I’ve bookmarked this podcast for later listening, but I’m curious as to why Exa is omitted by name from this post. While you do seem to be focusing on open source and Exa is commercial, their success as a publicly traded company seems to answer your initial question about betting on LBM to shake-up CFD. (You could place your bet in terms of buying their stock!) On the other hand, I don’t think you can say that Exa has shaken up the CFD world – they’re very much like every other player.

    1. Great question, John. In fact, (with the exception of one slip-in of OpenFoam), I intentionally did not mention the brand names of any CFD products. I wanted to focus on the current dance partner and thought it might be rude to ask him to comment on competitors.

      Exa is the obvious LBM gorilla- and there are some extremely smart people over there. Their strategy, however, focuses on very high end customers in specific verticals with high-end needs and deep pockets. That is a fine business strategy, but will impact a smallish subset of industry.

      I’ve also had exposure to NextLimit and XFlow, which is also promising technology on the LBM front.

      Matter of fact, I may do an interview with folks at those companies at some point too. One important piece to remember: If you have no competitors, it’s tougher to validate a separate category in the public mind. So, I’m a big fan of more companies pushing LBM.

      Flowkit may be in a position to have faster/wider reach through an intentional Cloud/SaaS model. Also, I think the OpenSource piece may pay great dividends. Or not. Gotta shake that magic eight ball again.

      1. I was referring to your writeup, not the podcast itself, regarding mentioning Exa. And regarding them, if LBM was really going to shake-up the CFD world I believe they would’ve done it already in the sense that if LBM was that great everyone would have migrated to it by now. Therefore, I suspect that your prognostication has to do not just with LBM but its combination with open source and “the cloud.” Is that accurate or am I still missing the boat?

      2. John, I’m just trying not to offend you by saying something like, “meshing is for suckers!” 😉

        I don’t think we’ve seen enough commercial work (on a large scale- with proper sales & marketing investment) to agree with the statement, “… it would have already happened by now.” Exa never had the public mindshare to compare with a Fluent, CFX, Star, etc. Just last year, I’d say “unscientifically” that over 75% of people in industry who I spoke with had never even heard of LBM. Those people had all heard of Navier Stokes and the traditional crew of tools.

        I think LBM has a lot of potential to shake things up on its own (so long as people focus on delivering in a way that fully takes advantage of the “mesh free” aspect). Having said that, it is always transient and in many cases more computationally costly than traditional CFD. So, a cloud/SaaS delivery model greatly helps to level the playing field for it. Even then, I don’t see traditional tools being swept aside. There will still be room for those in many applications.

        The OpenSource bit is (in my opinion) just an intriguing flavor added to the mix.

      3. Jonas Latt · ·

        If I may contribute to this exchange with my humble opinion, I think that John has a point when he mentions the central role of cloud computing to the future of CFD. But as far as the success of lattice Boltzmann is concerned, I believe that the cloud is an effect rather than a cause. Parallel hardware has become incredibly complex, lattice Boltzmann is successful because it knows how to use this hardware, and the cloud is successful because it remove the hardware from the user (I am simplifying, obviously).

        But then, the no-mesh approach is a crucial ingredient to computational efficiency, I certainly agree with Jeff on this point. The more successfully CFD solvers are parallelized on many-core hardware, and the more painfully you feel the effects of the bottleneck in the pre-processing stages (mesh generation and others) which scale more poorly.

        It hasn’t been such a long time now that we’re dealing with mass-produced many-core hardware, and CFD is (unfortunately) a field with a slow adoption rate for new methods. It makes sense to expect important innovations from lattice Boltzmann, even though it probably has not been an essential game changer so far.

        I’d say that we addressed these points to some extent in the Podcast; listening to it should clarify my standpoint. If not, we’ll have to record a third one I guess 🙂

  2. Ed Williams · ·

    Very interesting and informative. Thanks Jeff and Jonas!

    1. Thanks… and thanks for listening!