This is All Things ANSYS, a podcast from the technical support staff at PDT. Episode 52, an interview about technology with ANSYS CTO Prithviraj Banerjee and VP and General Manager of the Design Business Unit, Mark Hinsbow.
Welcome once again to our podcast. I'm recording this the day before I enter a turkey-induced coma from Thanksgiving, but you'll be listening to this probably after Thanksgiving. So I hope all of our listeners in the US had a nice holiday and enjoyed the food, the family, and friends, as well as any other activities you had over the four-day weekend. I hope everybody had a good time. This is a very special episode of the podcast, a little bit different. We're going to share an interview we did back in July. The original plan was to talk with four or five of the senior executives, VP and above type folks at ANSYS, but because of scheduling and conflicts and such, we were only able to do this one discussion with the two people that were able to make the initial interview. And we never could really pull off the things with the other executives. They were all very eager. We kept on setting up times and we kept on canceling. So instead of waiting another six months, we thought we'd go ahead and just share what we have, which is still a very, very good discussion. In fact, it's a 45-minute talk that we had with the CTO and the VP of the design products. So why don't we take a listen and I'll let them explain what they do and we'll have a great discussion and we'll see you on the other side. Hi there. We're here talking today with a kind of a who's who of the people that run the various business units at ANSYS, Inc. And I want to thank all of you there in Pittsburgh for joining me today here in Phoenix. Hi guys.
Hey Eric, how's it going?
So, there's quite a few of you there. So let's just start off by letting you guys go around the room and introduce yourselves and let us know your name, your title, what business unit you're responsible for, what your responsibility is at the company, because I know some of you are over the business units. And also, we always like to know how long you've been in the world of simulation.
Okay. Well, let me begin. My name is Mark Hinsbow. I am the VP and General Manager of our Design Business Unit. The Design Business Unit is here to get simulation out in the hands of every single engineer. You know, not just the analyst, although of course we love the analyst at ANSYS, but we love simulation so much that we would like every engineer to benefit from it, right? And we can talk more about that later. I've been with ANSYS for about four years and I have a broad software business background and I'm actually a nuclear physicist and wrote simulations in a whole different realm, more high-energy physics, way back.
Well, you know, I've always wanted a plasma simulation tool in ANSYS. So in your spare time, if you want to dust that off and put it in there, it'd be really helpful for our semiconductor customers.
Semiconductor manufacturing customers, yeah. And maybe some of our aerospace people as well.
Yes, starting to get into that plasma. Okay, who's next?
So I am Prit Banerjee. I'm the Chief Technology Officer of ANSYS and I joined the company about a year ago. Previously I used to be CTO at ABB, Schneider Electric, and before that I spent some time in academia. I worked in the area of simulation for more than 25 years. My role as CTO is to set the long-term technology strategy for ANSYS. So to go beyond the current products that we have, what should we be working on in terms of new technology, newer products, newer solutions and services. That is my role.
Very cool. You get to do the fun stuff. Let me go ahead and ask my next question. And I just saw Mark a few days ago, so some of these are actually spawned by some of the conversations we had when I was there in Pittsburgh. But one of the things that I kind of came away with from my last trip out there was there's some of these big megatrends that we're seeing out there in simulation like autonomous vehicles, electrification, what we see here at PDT is people trying to put more electronics in less space that generates more heat. What are you two guys focused on in terms of these kind of megatrends and what is ANSYS doing to help with those trends?
So let me begin and I'll have Mark chime in. So you're absolutely right. We obviously, as you know, we have a whole set of products at ANSYS on our mechanical products, fluids, electronics, and so on. But those are products. What we are working on is sort of solutions in front of our customers. So we are working on autonomous vehicle solutions, right? And so essentially when you're looking at AV, you have to look at all these multiple sensors that you have, the camera, the LiDAR, and the radar, and so on. And essentially we're bringing in our products, our solutions from our Optis business unit, our electronics business unit, and then putting it into what is called, we create scenario generation and so on. Then we are essentially putting out a complete vertical solution with our top customers. We just did an announcement with BMW to talk about ANSYS AV solution in the autonomous vehicle space. Similarly, we are working on electrification. Now many of our automotive customers are in the transition going from the internal combustion engine to the electric vehicle, right? So what is involved in electric vehicle, right? So essentially the electric motor is a changing thing, right? So we have our Maxwell product but trying to look at it, well, how will it actually drive electric vehicle. Then we have talk about batteries, battery modeling, simulation, so on. So we have a complete sort of vertical solution for electrification and we are working with the likes of Volkswagen. We just announced a very exciting partnership with Volkswagen and with COMSOL on our electrification solution. Moving beyond, we have identified sort of 5G as a big megatrend, right? Yes, companies like Ericsson and Nokia, they're all focused on 5G and everybody's excited about 5G. So we have a 5G set of solutions that we're working on which is all about antenna designs and how do you place antennas and how do you do sort of PCB designs to support antennas and all that kind of stuff from electronics, semiconductor electronics, and all the way around, right? Then in terms of other megatrends, I mean, people are talking about additive manufacturing, right? Big deal. So we have a solution for simulation to support additive manufacturing. Then we have a megatrend around sort of digital twins. Everybody talks about IoT and digital twins. And when I worked for ABB and Schneider Electric, we used to talk about sort of IoT-izing all our products. So we have a fantastic physics-based simulation tool to drive digital twins across all industries. So those are a few of the megatrends that I'm focused on. And again, Mark is focused on another very, very big megatrend around Discovery Live, which I'm asking him to chip in.
Yeah, and I think there are the solutions that you and I spoke about, Eric, right? I mean, we all see the same, right? Because we all participate in the same industry and they're really driven by our end customers and their needs and how they can innovate their products. And they're all the ones that Prit mentioned and maybe a couple more. And from a simulation perspective, the awesome thing is that these things have now become so complex that theoretically, you know, maybe 20 years ago, you could go address some of these challenges without simulation, but you just can't today. The, you know, 5G and the density of cell towers and the latency rates and everything else, right? I don't know how you would solve that engineering challenge in a reasonable go-to-market without simulating the whole thing. Similarly with autonomous, right? You have to drive billions of miles. And people have estimated that that's going to take us something like 8,000 years to do physical, right? And I don't think any of the car companies have an 8,000-year go-to-market plan, right? So you simply have to drive them virtually or simulate them virtually and so on. So that drives our business. And then we have some other trends that might take, you know, all of those are trends that take more simulation into classic product development if you want. And then we have a couple other trends, you know, like democratization where we say, hey, how can we help every single engineer make as many decisions as humanly possible themselves as they're iterating instead of having to round trip. It's almost like in the old days, right, where if you were in the accounting department, if you had to go ask the mainframe guy to run a program every time you have to have a new report, right? There's going to be less iterative insight. So that's one trend in simulation. And then you could argue digital twin is another one where it actually isn't the human being spawning the simulation anymore, but it's the machine spawning it on demand itself based on its sensor data and maybe an AI helping it. And then that will spawn a simulation to make a decision on maintenance or operations or so on, right? That's a whole, you know, different trend in and of itself as well.
Cool. Yeah, I think AI, we'll have to come back in a couple years and talk about AI because I think it's going to be a great discussion to have in simulation. But I want to go back to something that Prit said and kind of it was, you know, sometimes we get so buried in these things that I forget the obvious. We really have made a shift from providing point solutions that the users had to cobble together to come up with a full solution to really looking at that whole workflow and what are all the tools you need to solve a given vertical problem. And that really has been a shift in the last couple years. It didn't really kind of dawn on me till you spelled it out there.
So it's an important thing that's from ANSYS that we are actually working on. So from the CTO office, I'm driving two initiatives. One around AI/ML applied to simulation and sort of platform, multiphysics platforms applied to simulation. And Eric, you just nailed it on the head. I mean, essentially how do you take all these different point tools, right, and across these autonomous, electrification, right, if you're a customer like Ericsson or a customer like BMW, right, how do you actually take these ten different tools and make it all work sort of work together through the right modern APIs so they can call the right APIs from sort of application services, simulation services, and infrastructure services. That's how we are building our platform. But yeah, that's what I'm really excited about and that's the technology that's going to drive these new solutions, right? And tying it back to what Mark was saying is there's a lot of stuff people are trying to do, our customers are trying to do, that they couldn't get there without simulation. So we need these tools.
Yeah, we could talk about that forever, but let's move on. So one of the questions I get a lot from users, and you guys might be able to help answer it, is, you know, how does your team decide really what you're going to work on in upcoming releases? You know, what does that process look like? I think there's a lot of mystery out there on how a feature gets decided to go into a code or a whole new code gets written.
Yeah, and I mean it is a pretty big and important process, right? But I think it starts with, one is, yes, there are a couple things that go into a single release, but I'd sort of back out and say, hey, we plan, you know, multi-release, even multi-year, right? It starts, on one hand, it starts with all the trends we just talked about. So if we want to help our customers solve the autonomous challenge and be able to simulate eight billion miles, what do we need to do, right? And maybe we need to add capabilities that we, you know, that we didn't have before. Maybe we need to partner with someone, you know, like in the autonomous, someone needs to figure out how to create scenarios. It is not helpful that every time you need to create another, oh, this is a four-way intersection and this is a three-way intersection and this is how they do it in Pittsburgh and this is how you do it in Beijing, right? So to be able to simulate it, it has to start with someone has to create the scenarios, for instance, right? Which is something we've never thought about before. So you start on one end with, hey, here are these, you know, large opportunities for simulation to help product development and what do we need to do to deliver value. And then on the other end of that, you have direct feedback from your customers in existing production that says, oh, all of this is great, but I need feature X, right? Your multilinear plasticity model is not good enough in this instance, or, you know, you're lacking a set of materials, or whatever it could be, right? So you have, on one hand, you have very long-term strategic, on the other hand, you might sometimes have very short-term, very clear tactical needs. And then what we do is we go in, we're actually in the midst of kicking off that process. We start out really individual business unit by individual business unit planning what do we think has to happen over the next eighteen months or so and back that into individual releases. And then once we've done that, we go into a planning process that is more cross-functional. So what dependencies do we have on each other? You know, I might have planned on something, but it's dependent on someone else. And that also layers in, to your point, right, those cross scenarios. If we are to solve the autonomous scenario and that includes, you know, mechanical, fluids, thermal, electromagnetic simulation, what are the cross things we need to put together? And then ultimately, once we've been through, you know, both of those big vectors, right, individual BU narrowly thinking and cross-BU company thinking, it ends up being a feature map for the next three releases or so.
So let me complement what Mark talked about, right? So there's three levels of customer input that we get, right? So I, as a CTO of the company, I am routinely sort of involved myself in CTO-to-CTO level conversations, right? So I have sort of meetings with CTO at Ford, for example, right? Hey, Ken Washington, what are your challenges, right? John Luck, CTO at GM, right? So we have those CTO-to-CTO conversations in one-on-one meetings, in conferences, and so on. We were at the Paris show, talked to a lot of CTOs about what their needs are in the aerospace, automotive, whatever kind of space, right? That is one set of things. Second thing is we have a customer advisory council. For example, we, I mean, across BUs, we have many of the sort of, we put together about twenty, thirty customers that come together and collectively they provide input, right, to our leadership. But the most effective way that we bring in sort of input requirements is from our large ACE organization. As you know, Eric, we have very large, fantastic ACE organization and those people, these ACE engineers are working with customers throughout the world, right? So we have ACE people from North America, ACE people from Europe, ACE people from Asia. They come in, like for example, last week we had the electronics business unit planning meeting, right? Four days of planning meeting where these ACE leaders kept coming in from around the world, right? And say from an ACE perspective, right, from North America, this is what we are seeing as requests for our HFSS product for EBU, okay? So they may say, rattle off these ten features that they want, right? I want this, everybody wants this, right? So you take this list of things, clearly it's a resource issue, right? You cannot solve all of it. So you look for commonalities, right? What are people looking for? Oh, we're looking for this megatrend like 5G and so on. Oh, we really need to solve that 5G thing, how you position antennas and so on. So we put that into a, right, or in the autonomous case, there's a feature for sort of autonomous driver processing, ADP. Oh, we need to do that. Oh, we have this non-conformal mapping thing that we have to do for meshing. Oh, we need a tau meshing. So essentially you have this long list of features that the ACE engineers and leadership brings together to our BU leaders, okay? And then of course we have to say, well, we have just a hundred people, right? What can a hundred people do in the next eighteen months? So we prioritize and we come up with some plans. In these EBU planning meetings, we'll bring, as Mark was saying, the other leaders come in and, oh, okay, you need this from a cross-view perspective from EBU to talk to our mechanical business unit because it's the multiphysics part that is needed. So this process just got started from last week and now it'll continue through October, right? And it's one of the big NPI meetings. And then in October we say, okay, these are the resources we have in the company and collectively, I'm sure you know that we spend about two hundred million in R&D, about fifteen hundred R&D developers. So what are these fifteen hundred developers going to work on on our product features, release one, release two, release three in 2020? That's sort of the overall scheme of things.
That's fantastic. I think there's a lot of people that wanted to know what that looked like. And the ACE input for a lot of our listeners is critical because that's where they interact. I mean, hopefully we have a few CTOs that listen, but I think that the majority of the users do have interaction with the support team. And the fact that that gets into the loop and is an important part of setting what's needed is really critical. So lesson learned, if you want something, need something, don't just say, 'I need it,' or, 'This doesn't work.' Make sure you tell them why you need it and what the value is to you so they can get it back into the development team.
And the last thing you said, Eric, is super important, right? Because we get, as any product software product organization, right, we get a million feature requests. So it becomes the art of how do you say no to ninety percent of them and which ten percent are the right ones. So the more you can get it with a use case attached, right, the better it is for all of us. And by the way, the better the feature, you know, becomes at the end, right? Because physics is so wonderfully horizontal, but the way you apply maybe the high-frequency analysis in the autonomous realm might be slightly different than you do it in the 5G realm. So you really need that use case attached so that we can make good product decisions.
Very good point. Yeah. So kind of along the lines of new stuff and new features, ANSYS has partnered with folks, acquired folks, introduced new products quite a few recently. My favorite because of my background in mechanism design is partnering on motion. So we're working with a South Korean company on a really cool tool that allows us to do very sophisticated rigid body dynamics type stuff, which I love. I'm an old ADAMS user, so having it built into ANSYS is great. What's like for you two, what's your favorite new tool, either acquired or written or partnered with?
Well, I would obviously be biased, so I would say organically I feel like Discovery Live is my favorite tool.
It's the coolest. I'll say that.
Yeah, it's definitely the coolest. But if I sort of take that off the table, I actually think that the Granta acquisition we did is pretty spectacular. And it's spectacular because it'll help all of us, right? There's so much material knowledge and really, really good simulations require, you know, very good material knowledge. And then as you look at the sort of, you know, early prototype development where you also have to think about other things than just, you know, what's my Young's modulus or what's my viscosity and so on, right? You also have to think about things like sustainability, you know, price, manufacturability, and so on, right? Granta really has a wealth of knowledge that I think we've only scratched the, I know that we have only scratched the surface of how it's going to make things better across the entire board for all of us.
Yeah, we're going to focus on a few things from my perspective as a CTO of the company, right? So I mean, when you look at our users, right, and they're using simulation, so what are they looking for in simulation in terms of improving the quality of simulation? They want to improve runtimes, right? Some of our things take like a thousand hours to run and obviously we would like to speed things up. So anything that we do to speed things up is of value. They're looking at the accuracy. You can speed things up if you are willing to give up on the accuracy, right? And so we need to work on the accuracy and runtime tradeoff. Then we have to look at the stability, the reliability of this thing. Does it work all the time, right? I mean, maybe it runs sort of ninety percent of the time, but ten percent of the time these things just croak, right? So I mean, we tried to do meshing and it didn't do the meshing and then it kind of fails, right? And the ultimate, the final thing is the ease of use, right? So essentially across our tools we have put in a lot of fantastic features in terms of, for example, mosaic meshing that, you know, that came out in our tool, right? It's fantastic technology which enables us to do more things much, much faster. With mosaic meshing the speed goes up through the roof, right? Same thing with our tau meshing that we announced in our EBU product, right? I mean, there were things that did not work in those most complicated antenna designs, right, or cell phones, and now it does work, right? So essentially it's addressing the robustness issue. So these are some product features that we are very excited about in terms of cool technologies like tau meshing, mosaic meshing that enables things with respect to runtimes, right? We have always had the issue, right? We run our things on our workstations, right? And then if you want to move things up, you have to move to high-performance computing, HPC. But you look at HPC, how many people can afford to buy a Cray machine, right? You cannot. So therefore, you have to think about going to the cloud, right? Once you go to Amazon or Microsoft Azure, you have like ten million cores if you could afford, right? Run it on there. So our Fluent tools run in a very scalable way on two hundred thousand cores on a Cray machine. So always we have thought about, can we go to the cloud, right? And of course we have in the past worked with our cloud hosting partners like Rescale and Nimbix and so on. But this last June, right, in R2 release, we had this cloud-native solution I'm very excited about, right? From the desktop you want to do cloud bursting, you want to go to the on-prem data center, you could do it. You could go all the way to the Microsoft Azure cloud, you could do that. And doing that in a very interesting way tied with some of the newer business model that we've got around ANSYS elastic units and so on that I'm sure you have seen. I think the future is all around the cloud, right? People will do CAD design on the cloud, they do simulation on the cloud, collaboration on the cloud. So there's a cloud piece and there's a platform piece that we are working on to allow for PLM, etc., etc. It's technology called simulation process data management on the cloud. These are really, really good because it's less about running a particular simulation faster, it's about the whole workflow. Can I improve the overall workflow? And that will be done through the cloud, through the platforms, the stuff that I'm talking about. These are things that I am personally very excited about.
Yeah, I think, yeah, I'm glad you brought it up. I think it's going to change the way we do our job and make us more efficient. It's going to be a big, big shift when all that's up and working. Being able to solve in the cloud is pretty exciting.
Yeah. Think about how our productivity tools, like when I had a Word document, I send it to you, Eric, for edits. Then you send it to Mark, right? They're all emailing our Word files together. With Google Docs, we can simultaneously edit the file, right? And it's so much easier, right? I'm doing editing from my cell phone, iPad, workstation, etc. The same thing is going to happen in the CAD and simulation world and we just have to be prepared for it.
Yep. Be ready. Ready for it and have the right solution. That's so true. So one of the things we talk about at ANSYS is really as a goal, especially as we've kind of made the transitions with a new CEO over the last couple years, is this idea of really making pervasive simulation a reality. So that's, you know, getting tools into the hands of more engineers so that more people can drive designs with simulation. So this...
This is more of a philosophical question, really, but what challenges do you see in spreading — not necessarily that we have to write more code — but what organizational challenges or process challenges do you see in spreading simulation deeper and wider in various industries?
I think, Eric, you and I have talked a lot about this. If you look at all technology that has become pervasive, it's become pervasive for two reasons, probably, mainly. One, you get ease of use, where a much, much broader set of the population could use it, right? That was true for the GUI on the PC and so on. And you also get to a certain speed, right? If we sort of go back to Gutenberg's printing press, that was not easier to use than writing, but you didn't take a monk a full year to write a book, right? You could do a copy in a couple of seconds.
And it was true to a certain extent. You could argue for the internet back in the day where all of us had to use dial-up modems and it took us 25 minutes just to get connected. You wouldn't check the weather or your stock quote, right? Because investing 25 minutes in that process made no sense, right? So you have to get technology to a place where it's reasonably interactive and where it's reasonably easy to use.
And then in terms of culture, on both sides, you also have to let go of certain things. And what you know on the inside — like, we are proud rocket scientists, right, in the broadest sense at Ansys — but that doesn't mean that we have to surface every knob and dial to the end user, right? And sort of, 'hey, look at how smart we are, right, here are all the things you could go tweak.' That's culturally not helpful.
You know, I a lot of times with my team use Google search, given that Prith brought up Google as an example. You know, being able to query the entire web in 100 milliseconds and serve up a result, I think all of us can inherently look at that and say, 'man, what a set of database query AI technology must that require.' But the user interface is a simple text box, right? Where it's not an SQL query where I do some boolean logic and choose database fields, right? In which case, Google search would have gone nowhere.
So part of us is taking the stuff that we're most proud of, right, the simulation technology, and to a certain extent hiding it. You know, you should not know, depending on Reynolds number or whatever it is, which turbulence model to use, right? It should just be slotted in on demand. You should not know which meshing algorithm — not the normal engineer should not know that, right, necessarily. You should be a good engineer, know about material properties and so on. But the fact that double-precision points have some error and we're doing numerical corrections on the back side should just be hidden. And that's probably internally at Ansys one of the biggest cultural things, because we're so, in a good way, proud of the stuff that we're doing and we love to be geeks.
But, you know, one of the things we just shipped, for instance, in the new Discovery Live is thermal stress, right? Everyone who's been in simulation for a while knows that that's multiple, you know, solvers being coupled to each other, but the setup is just thermal stress, right? It's, 'Hey, I want to do some stress analysis.' Fine. You know, give me some kind of temperature or heat input and give me some structural boundary condition and we're done, right?
So that's one. And then of course on the customer side, there's an equal cultural change that has to happen, because processes will change. One of the things I preached recently, and that got people's eyebrows to pop up a little bit, I preached what I've provocatively called careless engineering.
