Tech Soft 3D Blog

Beyond 3D Podcast: Simulation and Collaboration in the Cloud

Posted by Tyler Barnes on May 22, 2018 10:19:16 AM

Simulation is something that anyone in manufacturing is familiar with, and in this episode of Beyond 3D we talk with David Heiny, co-founder and CEO of SimScale.  David talks about how simulation in the cloud has evolved, how the cloud has enabled more powerful and rapid simulation, and also where simulation is headed – that it is much more than just the power of the cloud, but how real-time collaboration simulation is now possible, even more accessible and that’s enabling more small and mid-size companies to take advantage of simulation early on in the design process.

Listen in and give us your perspective.  Are you doing simulation? Or have you wanted to but never had the resources?  

To learn more about SimScale, visit www.simscale.com

To read the full podcast transcript, see below. 

read-transcript.png

Angela Simoes:

Welcome, everybody, to another episode of Beyond 3D. My name is Angela Simoes, and we are here to talk about simulation today. Our guests today are Gavin Bridgeman, chief technology officer of Tech Soft 3D. Hi, Gavin.

Gavin Bridgeman:

Hi.

Angela Simoes:

And our guest today, our expert on simulation, is David Heiny, who is co-founder and CEO of SimScale. Welcome, David.

David Heiny:

Hi, Angela. Thanks a lot.

Angela Simoes:

Yes, thanks for being here. So let's just jump right in. Tell us a little bit about yourself, David, and SimScale and how the company came to be.

David Heiny:

Sure. I, myself, am an engineer and mathematician by training, and a couple of years ago, basically during grad school, me and my co-founders ... We're five founders here at SimScale, all of them having a background in engineering or software engineering, and we basically founded a small engineering service provider in parallel to the grad studies. And what we've done is basically we've been using Cloud technology from the very beginning. So we ran inside of the Amazon Cloud our simulations, provided services on our own simulation software stack, partially open-source codes, provided services to mainly German companies, classical engineering services.

 

During that time, our customers back then asked, "Oh, that's an interesting approach you're taking. You guys don't own any hardware and just use Cloud resources whenever you need them, so it would be great if we could get access to that or could leverage that as well." And this was sort of where the idea sparked and we said, "Okay, hey, what could be an easy way to deliver to our customers the tools we're using?" And so we looked into how could we deliver that via the web or should it be a remote client? And this was sort of the early days of SimScale, I guess, and the first lines of code written.

 

And that evolved into what SimScale is today. We discovered that the Cloud ... and I think we're gonna talk about that later a little bit, that the Cloud is much more than just computing power in there, but if you think it through and if you look at how the simulation market works today, how the customers are served today with simulation solutions, we believe that there is a big potential, and that's when we started it. Nowadays we're headquartered Germany, roughly 50 people, so it's still early days. But yeah, so that's maybe a few words for how things started.

Angela Simoes:

Sure, and so it's interesting that you said you had been based in the Cloud from the very beginning, and simulation has certainly been around for a long time. So from when you guys got started to even now, how has simulation evolved over the past few years? Are people doing new things, different things? Or, we talked about it's more than just the computing power of the Cloud. Maybe talk a little bit about how simulation itself has evolved, and Gavin, chime in as well if you'd like.

Gavin Bridgeman:

Sure. I think one thing, David, I'd be interested in ... I've used your application. I don't come from the background of simulation, but I found it very easy to get started with setting up a simulation in your system. Is that something that's standard in other applications, or is that something that you brought that was a focus on this initial release of your application?

David Heiny:

Good question. I think the accessibility of simulation in general, not just our application, has improved a lot over ... Simulation is a couple of decades old, and I think it came a long way since then. So people solve more complex problems, multiphysics problems, larger problems, and I think this is an ever-increasing trend into that direction while the accessibility, the usage, or how intuitive these applications are has also improved a lot.

 

But you're right, Gavin, that our focus or our belief is that still today if you look at the number of engineers or the fraction of engineers that do regularly use simulation, that use it to its full potential is marginal compared to the people or engineers being involved in technical product development. So we do think ... and depending on which market research company you believe in, they believe it as well, that there's still a big gap when it comes to enabling engineers to use simulation.

 

And you're certainly right that this is what we're passionate about here at SimScale: so surely, removing these barriers that are still around that prevent engineers and designers to leverage simulation, to remove those with the help of the Cloud by porting an entrant simulation workflow to a SaaS model.

Angela Simoes:

So you mentioned something about that use of the Cloud is not just about the computing power. Can you go into a little bit about what you meant by that?

David Heiny:

Sure. I must say that it's also been a learning journey for us at SimScale. When we started, for us, it was mainly ... We were using these tools without the user interface. So everything was Linux shells, and we've used these codes simply by a terminal, and so it was just about, "Okay, where can we get computing power on demand?" And so, for us, it was in the early days really just about the computing power.

 

But over time, we've understood more and more that this is certainly an aspect that's important for simulation, because depending on what you're doing ... For simple, linear, elastic FEA simulations, it's certainly not a big problem. You can run that on your laptop, but when you're looking at more complex, non-linear structural mechanics or also going into fluid flow, then you're looking at significant computing consumption. And computing power is certainly a good thing, but the true change or, I think, the disruptiveness of putting a simulation application into the Cloud is that you have a completely centralized stack.

 

Quickly, as a background, at SimScale, we're running on AWS. It's a Web application, so all users simply access the application or web browser, and everything runs server-based inside the Cloud. So everything is persisted there, so nothing is persisted locally. And so this allows us to change how simulation is being consumed by users, and that starts with obvious things such as collaboration.

 

You can real-time collaborate on simulation settings, and that could be setups, and that could be within an organization, a simulation expert with maybe a design engineer that is not using simulation so often. That can be our support with our customers, and that could be a consulting engineer with his client. So being able to real-time collaborate on engineering projects and simulation in particular is a big asset. And with, I'm sure, in CAD it's a big thing. There's PLM systems. There's lots of engineering applications that are nowadays being made more collaborative, so that's one thing.

 

But then if you think it more through, it starts with things such as data science at the backend, things as we have the ability to help the user in a different way because the analyses we can run on the backend to see, "Oh, why haven't a particular set of simulations not been successful?" And so we could jump in there and help the user and maybe identify if there's an insufficiency in our product and improve that, etc.

 

So I think it's more than just a computing power. It's really the change from a desktop client where the software vendor, the customer, the support team all operate in silos separated from each other to a centralized stack, a collaborative system where everybody works on the same application in the same environment, and you can turn or use this fact to make the application better for everybody in there.

 

With the danger of talking too much for this answer, maybe one very crisp example to showcase the power of such an approach is, we've introduced a while ago a free plan on SimScale so that you can use it entirely for free with all features and with significant computing power for free. But in that community plan that comes free of charge, the limitation is that you're only allowed to create public projects. So all of these projects are publicly accessible, and other SimScale users can see them and can reuse them for their purposes.

 

And in the first year after introducing this new plan, the success rates of new users ... I think it was like 3x or 4x, so they grew so heavily simply by the fact that users could leverage simulation projects of other users that are similar to what they want to do in a collaborative manner. And these are things that are simply not possible in a desktop environment or possible with so much friction that adoption is small.

 

And I think that's just a few thoughts on why we think the Cloud is not just computing power. It's more. It's collaboration. It's data science. We could also talk about pricing. And I think that's the potential.

Gavin Bridgeman:

One interesting part there that you're bringing up, David, there is this idea of seeing what your partners are doing with your software. From that perspective, that's one of the fantastic things that you have with Cloud software. You really have the ability to see what your users are using. From how that's impacted where you've grown the software, has that impacted you, for example, investing in certain solvers or has it been around usability issues? And what kind of impact has the ability of seeing what your customers are doing with your software impacted your product development?

David Heiny:

That question also has the danger that I talk too much, Gavin. But trying to keep this crisp, the short answer is, heavily. I think, again, in the desktop realm, it's not that you wouldn't have any information about what customers are doing with your software. There's crash reports. There's certain usage reports, etc., so it's not that you do not have anything in the desktop realm, but you do not have it in real time. You do not have it in that depth and that width you have in the Cloud, and so, it obviously heavily impacted what we were doing. And it starts with, yeah, observing how successful users are. And if they are not successful, why are they not successful? Does it require more training? Is the onboarding not good enough? Where are people dropping out? Is there a certain piece in the workflow that's not discoverable? These sorts of things and these classical product management questions.

 

But then also, obviously, in the simulation space, when you're looking at more complex simulations, where it's really non-linear problems, where you have iterative solvers, you can start looking at, okay, how do certain solvers converge as compared to others? Should we change default setups? Sorts of things. So just the fact that you can deep dive into these things and use it to make the product better for the customer is, I think, something that has heavily impacted our product development.

Angela Simoes:

I wanted to just jump back to a couple of points you were making before, because it sounds like you were really talking about real-time collaboration and then using the Cloud for other things throughout the design process. So you're talking about doing simulation throughout the entire process and not just at the end, and is that something that has changed as simulation has evolved? And it also touches on the topic of mindset. Does that require a change in mindset that you can do simulation as a larger team and now collaborate versus make your design and then simulate at the end to validate what you have? That you're now doing simulation more thoroughly?

David Heiny:

Yeah, certainly, and again, that's not a trend that we at SimScale own or we are the first ones to push into that direction. Rather, I think the simulation has been due to the hardware requirements, due to the complexity of software applications in the early days, and that PhD that needs to sit in front of that software to make effective use of it using this high-performance computing system that is ... I don't know ... tens of thousands if not hundreds of thousands of dollars investment. Requiring these huge amounts of computing time was just something you better think twice before you run a simulation, so when would you use it? When it's inevitable, and that's by the very end, when you try to run a final validation or for products where ... in the automotive or aerospace sector where the investment into software stack has been huge compared to other verticals.

 

So this is where it came from, and now with the simulations coming from an economic perspective but also from a know-how perspective more accessible, I think people start using it early in the design process, in other industry verticals, where it was before not economically feasible, in settings where not just the analysis engineer or the scientific computing engineer is using it but also the designer can make safe use of it.

 

And we still think there is still lots of work to do, because from our perspective, the situation today is still that there is a big trade-off still around today where you need to decide for either cost efficiency or high-quality simulation results. It's still something where this trade-off is there. And I think that has been around in the early days also in 3D CAD, but today, I think this difference in 3D CAD is different than it is in simulation. You'll get affordable, highly professional 3D CAD packages today for a good price that is affordable to any kind of engineering company. But in simulation, it's still different. And also, from a know-how perspective to make effective use of that, it's still different. And so that's what we're trying to tackle, to get this know-how barrier down, to get this budget barrier down to simply democratize it further.

Angela Simoes:

And so then that brings up the question of ... 'cause you mentioned aerospace and automotive, who, as everybody assumes, has really large budgets to invest in something like this, but now that simulation is becoming democratized and more affordable, what other industries do you see picking up doing more and more simulation? And what kinds of simulation are they doing?

David Heiny:

It might be sort of also a self-fulfilling prophecy, because at SimScale we're focusing on the mid-market, I would say, so smaller, medium-sized companies. We do have also some larger companies, but it's rather the industries and the type of companies that are already further down the journey of accepting a Cloud solution as an engineering software in their stack. And what was particularly interesting for us coming from the mechanical engineering background was that we've seen a lot of ... So meanwhile, the architecture space, the AEC, is one of our fastest-growing verticals on SimScale, and that was very surprising because it was not us saying, "Hey, we could try that." It was really the other way around. On the SimScale community, we saw more and more interest in this. We saw more and more public users, also professional and paying users to run these types of problems in the architecture space. And that's how we discovered that and so we've been basically developing particular features for that industry. So that is one simple example.

 

But obviously there's also more traditional verticals, such as we see a lot of electronics coding at SimScale as well, so thermal management of enclosures, pumps, valves, also lots of UAVs, and yeah, these sorts of things. And obviously classical product design, where you're looking at traditional AEC, that's also there.

Angela Simoes:

What's so interesting about AEC is that ... Are they simulating things like airflow through a building or, I guess, the load for a frame? Or what are some of the things that they're looking at?

David Heiny:

Yeah, what we're seeing at SimScale is mainly flow simulation, and that starts with city-level flow simulation, so we're really looking at urban wind comfort and then over to a single building, where you're interested in how can you ventilate your building without using a mechanical system internally? So how could you naturally ventilate the system? And then inside the building, looking at thermal comfort analysis of large-occupant spaces. It was really interesting to see that, and then down to more specific problems such as contaminant control inside parking garages, these sorts of things, and also I think it makes sense if you think it through because there's more and more energy-efficiency requirements in building codes globally in every country. The regulatory requirements around buildings get more restricted, and so the need for civil engineers or sustainability engineers to iterate fast on building designs that ultimately meet these requirements, it makes sense. It's sort of a global trend, and the building industry is heating up. So I think from that perspective, it makes sense, but it was certainly not that we were smart enough to see that upfront. It was rather that they came to us.

Angela Simoes:

And when you have conversations with partners and customers, when it comes to simulation, what things come up in conversation? What are some of the questions or concerns or maybe even challenges that customers are having that you address with them when it comes to simulation?

Gavin Bridgeman:

One of the interesting parts that I’d like to get David's perspective is certainly we heard in the US, there's obviously a big momentum for the Cloud, and people have really embraced the Cloud, while when we spoke to a lot of German-based companies, there was a lot more reticence around moving towards the Cloud, so that would be one interesting thing to hear David's perspective on. Certainly you look at the Cloud and people intuitively understand that simulation is a computationally intensive operation, and consequently the Cloud brings that powerful computing to the table, so people see that's the real value there. And so it'd be interesting to get David's perspective on the huge manufacturing industry and a huge need for simulation, whether the obvious strengths of the Cloud, that audience was more open to using Cloud technology than what we've experienced with other German-based companies.

David Heiny:

Yeah, you're certainly right there. To me, it's oftentimes fascinating ... I'm not sure, Gavin, if you have the same experience, but sometimes when I talk to people outside of this industry that are not in the engineering software space but in another software space, when you tell them, "Hey, we're working on a SaaS solution for simulation," or I'm sure other founders or software people, they're building a SaaS solution for another type of engineering software have the same experience that they answer, "What? Is that still a thing? Hasn't that been already done? SaaS?" Because SaaS has happened in every software vertical, and it seems like engineering software seems to be one of the last verticals where it hasn't happened yet, and I think for a reason. It's very complex applications. It's graphics-heavy. The customers that are using it are demanding they have an existing overflow in place where they've built a lot of IP based on.

 

And so I think we are facing that challenge from the very beginning. We are convinced that the future will be in the Cloud and that lots of engineering work will be done in the Cloud for advantages and for reasons we've already talked about, because they are so dominant these advantages. But it's gonna take time, and certain countries and certain industries will be faster than others.

 

And I agree with you, Gavin, that we're seeing that the US is our strongest market and that there, I think the mentality and also I think ... I don't have any data, but I could imagine that also in other SaaS solutions or other software verticals where SaaS applications popped up, it might be also rather an early adopter country compared to others. And I think we embrace that challenge and know that we cannot alone educate this entire market. It's many software vendors. It's the existing desktop software vendors that start offering Cloud solutions.

 

So I think over time, that level of acceptance will rise, and I think that's how we manage the situation, simply that you cannot change the fact that some verticals will never move to a public Cloud application because of certain regulations. Some of them, it will take them years, and others are ready today, and I think that, particularly as a startup, that's a reality to face, and that's how we operate SimScale today.

Gavin Bridgeman:

And Dave, another thing that ... As I mentioned, I played around with your application, and it's a beautiful application. I love to see ... and simulation is such a complicated subject with respect to all the information that you have to display and you can play around with, and it's great to see 3D integrated into this very sophisticated Web application. So I'm assuming you're using WebGL technology, and I'd be interested to talking a little about these different technologies that you use behind the application, and obviously there's WebGL, which allows users to get to Web, 3D and the browser. There's also the GPU on the Cloud technology that's now part of the Amazon stack. Could you talk a little bit about those different 3D technologies that you're using in SimScale?

David Heiny:

Sure. Basically the most of the 3D interaction on SimScale is ... Let's start the other way around. You're right. We're running almost everything on Amazon AWS, and the good amount of 3D interaction on SimScale happens via WebGL, so clients had rendering on the GP of the client, and we're using that for parts of the workflow where significant interaction is necessary and where the model size is manageable for the client, and that's particularly the case for CAD models as well as partially the meshes that are generated on SimScale, because there, the data sizes are manageable still for the client.

 

Opposed to that, once a simulation has been carried out ... Let's say a finite volume simulation has been carried out on, I don't know, 20 million cells, and you now want to interactively slice and cut sections through these 20 million cells, and you want to visualize that cut section and then show the velocity field on that cut section, that's such heavy data that needs to be processed that we're using a service-side-rendering mechanism for that, and in that scenario, we're leveraging GPUs inside AWS, where these operations are carried out on and where essentially pushed down to the client.

 

So it's sort of there's additional challenges, but we're always trying to use best of both worlds and are using the 3D technology that suits the particularly step in the workflow the best.

Gavin Bridgeman:

Right, right. Yeah, it looks fantastic, and it's really responsive. My simulations were only up to scale that I could use the WebGL side of things, but you mentioned using the GPU on the Cloud, which makes perfect sense, and it's fantastic that that's connected in.

 

Another part, David, that's kind of interesting here was you're mentioning to me around that your release cycle is ... You're releasing your software on a daily standpoint. For an application that's sophisticated as yours with 3D incorporated, how does that look from a testing standpoint and what you do from it? You must obviously automate a bunch of tests, but I'd be interested in talking a little bit about how that works and if releasing software on a daily standpoint really changed obviously investments, but also how you developed software.

David Heiny:

Yeah, I guess there's many disadvantages with building a new simulation and infrastructure and architecture from scratch, but one of the advantages is that you can also think freshly about certain aspects of how you want to develop that piece of software. And so, I think from the very beginning, we've made sure that we're using modern best practices around modern Cloud development. So I think while on the one hand, we are certainly having similarities to how other simulation software vendors are developing their software, automatic validation, large benchmarks being carried out automatically, etc., but then, we're also having lots of best practices that you can, to the best of my knowledge, only find in a Cloud environment, where we're talking about continuous integration, continuous delivery. And so, this is how we've built our entire stack, that automation is the default for everything.

 

And so if you really want to be able to deploy whenever you want, to release whenever you want, and yeah, on average, on a daily basis, you need to make sure that this is actually safe for the customer. And so, on a constant basis, our automatic validation benchmarks, our automated tests are running, and they are ever increasing, and yeah, that's sort of the ... If you want to take advantage of the fact that you actually can work so closely with your customer and fix things as fast as possible, ship things when they are ready and not whenever your release cycle ends or starts, then you need to build your development infrastructure and your development process around that as well.

 

I think it would be too much to go too much into detail in that conversation here, but I think, yeah, the fundamental principles are automation first and lots of automation built into our stack that ultimately allows us to ship on a daily basis where you can find many parallels also in other Cloud applications and how they are developed.

Gavin Bridgeman:

And do you see much as a platform that has a equity-sophisticated application, and that includes 3D, do you see many challenges that you run in with different browsers running on different operating systems or different hardware? Or are you finding that's a fairly stable platform to build on top of?

David Heiny:

Yeah, I think it all comes with advantages and disadvantages, but I think we do not have many challenges that some software vendors have, where at SimScale is at its core a cloud infrastructure, so we basically do not deal with, "Oh, we need to build something for a particular target operating system," or something. Cross-browser testing is also something that can meanwhile be automated. There's also a bunch of manual work we still do, but cross-browser testing is something where there's frameworks around where there's other companies, even startups, offering solutions around that. So lots of these things can also be automated, and again, I think that's nothing we invented or something but rather because so many sophisticated applications are meanwhile delivered via the Web browser, many other companies face the same need or have the same need, and so, naturally, solutions pop up for that, also tons of open-source solutions, etc. So that's really something actually we see as an advantage rather than a challenge at that point.

Gavin Bridgeman:

No, that's great. And I asked a lot of questions, David. A lot of our people who have relationship with Tech Soft from a technology provider perspective, they're from the traditional desktop environment, and so I wanted to just hear you talk about some of these things of deploying a modern kind of Web application and how some of the challenges relate to that.

David Heiny:

Yeah, and just maybe to give two crisp examples. I think a challenge is certainly the fact that there is no time to do any manual work if you want to ship something. So you need to rethink and make sure that your entire stack supports such a deployment, 'cause that's certainly a big investment. That's something that takes time, and that's certainly a challenge.

 

I think one advantage where we're seeing is really that, let's say, if there's a bug fix or a new feature, a bug fix, you can even go down and see what exact simulation run was affected by a certain bug? And you can help your customer troubleshoot that bug fix in a much more effective manner and in a much more convenient manner for the customer, where in the desktop, you would be sort of left alone with that. A feature, same thing. If a feature is ready and it went through QA and it's actually release ready, we can ship it right there, and it can right away serve a customer instead of waiting for another six months until you can ship it.

 

So there's disadvantages but also advantages in that.

Gavin Bridgeman:

Great. Thanks.

Angela Simoes:

Okay. Well, this has been a pretty awesome conversation. I think we got into some pretty good detail and hopefully answered some questions that our listeners have about simulation. I would just say, if there's any final thoughts you have, David, about if customers have maybe done it and are stuck or maybe they haven't done simulation and are thinking about it, what are some of the things that you tell people when they're looking at simulation capabilities and options and things like that, how to get the most out of their investment?

David Heiny:

Angela, my answer might be biased now.

Angela Simoes:

Of course. But that's okay.

David Heiny:

I think the barriers of just trying it today are almost nonexistent anymore. I, myself, I'm using many different engineering software solutions completely browser-based. While the vision of an entire engineering software stack and a browser hasn't completely happened yet, so some parts of the workflow are not there yet, most of it is already there, and it takes 20 minutes to give something like this a spin, and what I oftentimes say is, "Don't take my word for it. You can just sign up." And most of these applications, and SimScale as well obviously, support free trial or even, in our case, a free community plan. So it's just something you don't have to take anybody else's word for it, but just give it a spin yourself.

Angela Simoes:

I would agree. Gavin, any last thoughts?

Gavin Bridgeman:

Just really appreciate some of the insights there, David. It's really interesting to hear, and I certainly encourage anybody to if you're just interested in seeing a very modern Web application, touch your toe in the water with some simulation software, SimScale works really, really well. Thanks.

Angela Simoes:

Excellent. And with that, we'll wrap it up, and thank you, David, so much for your time today and your insight. I think it was really helpful, and I'm pretty sure our listeners are gonna feel the same way. Thank you also, Gavin, for your time, and for those out there listening, thanks for spending some time with us today.

 

And if you haven't hit Subscribe yet, please hit that Subscribe button on SoundCloud or on iTunes, and please share this podcast with your colleagues and other folks that you know that might be interested in this topic or any topic beyond 3D, if you will. And if you have a few minutes, please leave us a review on iTunes. It actually does help others find our podcast much more easily, and believe it or not, it actually does help. We're not just asking for some kudos. We're asking for your help in getting others to be able to find us, so thank you, again, all around, and until next time, have a great day. Thank you.

 

Topics: Cloud, Simulation, Manufacturing

Subscribe via E-mail

Latest Posts

Follow Me