WEBVTT
00:00:00.000 --> 00:00:02.879
When you do this at home with your plastic printer, not a big deal.
00:00:03.040 --> 00:00:07.440
It takes a couple of minutes, you know, to reset the printer and to start printing the process.
00:00:07.599 --> 00:00:24.800
But nowadays, when we look at metal, industrial metal additive manufacturing printers that print very large parts, some to you know, up to about a cubic meter of build volume, where they print rocket parts or very large parts for the defense industry.
00:00:25.519 --> 00:00:38.399
If a failure occurs when you print such a large part, it's a huge waste because chances are very, very slim that you can restart this printing process and continue when this failure occurred.
00:00:38.560 --> 00:00:42.000
So, more likely, what you have to do is you have to take everything out.
00:00:42.159 --> 00:00:48.880
It's basically considered scrap, and you have wasted time and money to try to build a part.
00:00:49.039 --> 00:00:50.880
Now, this is where simulation comes in.
00:00:51.119 --> 00:00:53.200
Welcome to the Additive Advantage Podcast.
00:00:53.359 --> 00:00:54.640
I'm Sean Anderson.
00:00:54.960 --> 00:01:01.520
This is where we explore people, technologies, and strategies that are shaping the future of advanced manufacturing.
00:01:01.920 --> 00:01:04.239
My co-host Danny Mason and I.
00:01:05.200 --> 00:01:06.879
Yeah, I'm sorry, I don't know what.
00:01:07.040 --> 00:01:09.040
Clearly it's not 9 30 at night.
00:01:09.120 --> 00:01:09.760
Yeah.
00:01:11.439 --> 00:01:15.359
I was reading two paragraphs at once and just kind of randomly mixing them together.
00:01:17.840 --> 00:01:19.599
That's how nope, this is all good.
00:01:19.760 --> 00:01:22.159
That is how my brain works all the time.
00:01:23.680 --> 00:01:31.599
Welcome to the Additive Advantage Podcast, where we explore the people, technologies, and strategies that are shaping the future of advanced manufacturing.
00:01:31.840 --> 00:01:34.959
I'm Sean Anderson, here with my co-host Danny Mason.
00:01:35.280 --> 00:01:45.519
Whether you are leading an additive program, evaluating technologies, or just looking for practical ways to improve manufacturing outcomes, we're glad you're joining us.
00:01:46.159 --> 00:01:56.239
Today's conversation tackles a topic that often flies under the radar in additive manufacturing, but arguably has one of the biggest impacts on production success, simulation.
00:01:56.480 --> 00:02:03.599
If you've ever had a metal print fail after hours or days of build time, you know how expensive trial and error can become.
00:02:03.760 --> 00:02:07.359
But what if you could identify those failures before ever pressing print?
00:02:07.599 --> 00:02:11.840
Joining us today is Christian Rossman, founder and CEO of Notavek.
00:02:12.080 --> 00:02:23.680
Christian has spent more than a decade at the intersection of metal additive, simulation, and predictive engineering, including time at Materialize and founding Additive Lab before launching Notavek.
00:02:24.000 --> 00:02:36.960
In this episode, we explore how simulation is evolving from a tool reserved for large companies into something accessible for manufacturers of every size, and how AI may be the catalyst that finally brings simulation into the mainstream.
00:02:37.120 --> 00:02:37.599
Let's dive in.
00:02:52.159 --> 00:02:54.080
All right, so my name is Christian Rossmann.
00:02:54.159 --> 00:02:55.199
I was born in Austria.
00:02:55.280 --> 00:02:56.800
This is where I grew up.
00:02:57.039 --> 00:03:02.479
Um and I followed an academic career for quite some time.
00:03:02.560 --> 00:03:12.479
So I studied mechanical engineering, um, did a master's in business engineering, and um got my PhD in computer science.
00:03:12.639 --> 00:03:20.159
And um that PhD was partially completed in um at the medical university here in South Carolina.
00:03:20.240 --> 00:03:26.800
So this was the first time I actually moved to the US and uh got to experience what life in the US is like.
00:03:26.960 --> 00:03:35.280
Then in 2014, my wife and I we moved to Belgium, and when we did that, I was actually looking for a job.
00:03:35.439 --> 00:03:40.800
So traditionally, my experience was in the traditional mechanical engineering field.
00:03:40.960 --> 00:03:53.120
But in 2014, when we moved back to Europe, I got a job at this company Materialize, which I'm pretty sure most of your audience uh know this company.
00:03:53.680 --> 00:04:08.639
And there I um got exposed to metal additive manufacturing, more specifically, how you can use numerical tools, predictive tools to um investigate what's going on in the printing process.
00:04:08.879 --> 00:04:11.599
And I'm assuming we will talk about this a little later.
00:04:11.759 --> 00:04:17.279
So there's a lot of things that can happen that result in failures which cost companies money.
00:04:17.519 --> 00:04:28.000
Um and so yeah, I got exposure to that, and we started investigating different processes um uh and different methods and how to do these predictions.
00:04:28.160 --> 00:04:51.519
And then in 2017, 2018 is where I found my first company, Additive Lab, um, in Belgium, that um was basically a research tool with um a variety of um numerical tools and applications to do to answer questions that one may have when it comes up to the metal additive manufacturing process.
00:04:51.600 --> 00:05:00.079
And in 2022, I my wife and I and my my family with our child, we moved back to the US.
00:05:01.199 --> 00:05:13.199
And this year, with having you know more than 15 years of experience in the additive manufacturing simulation field, specifically on the process prediction side.
00:05:14.560 --> 00:05:34.319
Um, I felt a need there's I felt there's a market need for a new tool with um a lower a lower cost burden that allows users with lesser budget to utilize the advantages of simulation and the predictive power to optimize and improve their metal additive manufacturing processes.
00:05:34.480 --> 00:05:35.120
It's wonderful.
00:05:35.279 --> 00:05:38.639
What a what a robust history too in the additive field.
00:05:38.800 --> 00:05:41.199
And you've touched so many different facets of it.
00:05:41.360 --> 00:05:50.160
I think what would be helpful, um, because our audience, some people are familiar, you know, with why you would do pre-print analysis or simulation if they have done metal additive.
00:05:50.240 --> 00:05:52.399
A lot of folks maybe have less familiarity.
00:05:52.639 --> 00:06:00.480
For someone who is unfamiliar with this process, what types of things can a simulation tell them before a part is ever printed?
00:06:00.720 --> 00:06:03.199
Yeah, so to me, I'm very passionate about it.
00:06:03.279 --> 00:06:04.959
I'm gonna be very honest with you.
00:06:05.199 --> 00:06:13.839
I do not understand quite why not everybody with a metal 3D printer is using predictive simulation software.
00:06:14.000 --> 00:06:36.000
So when you have a metal additive manufacturing process like laser powder bit fusion, direct energy deposition, um sinter-based processes like metal binder jetting or cold metal fusion, all these processes, there's a lot of complicated physics happening that can lead to manufacturing failures.
00:06:36.240 --> 00:06:44.240
So everybody, even with a with a desktop 3 printer at home, I'm sure has experience you start the printing job and then all of a sudden the job fails.
00:06:44.319 --> 00:06:46.000
So you have to redo it all over again.
00:06:46.079 --> 00:06:48.959
When you do this at home with your plastic printer, not a big deal.
00:06:49.120 --> 00:06:53.519
It takes a couple of minutes, you know, to reset the printer and to start printing the process.
00:06:53.759 --> 00:07:10.879
But nowadays, when we look at metal, industrial metal additive manufacturing printers that print very large parts, some to you know, up to about a cubic meter of build volume where they print rocket parts or very large parts for the defense industry.
00:07:11.600 --> 00:07:24.480
If a failure occurs when you print such a large part, it's a huge waste because chances are very, very slim that you can restart this printing process and continue when this failure occurred.
00:07:24.639 --> 00:07:28.079
So, more likely, what you have to do is you have to take everything out.
00:07:28.240 --> 00:07:34.959
It's basically considered scrap, and you have wasted time and money to try to build a part.
00:07:35.120 --> 00:07:37.040
Now, this is where simulation comes in.
00:07:37.120 --> 00:07:51.040
So, what simulation does is typically in form as a uh of a software tool that takes in, you know, for the most part, a couple of minutes, maybe up to an hour of time, that you load into uh your geometry.
00:07:51.199 --> 00:07:59.519
So when you prepare your build, typically some 3D geometries, you run the simulation and it will tell you if there are certain problems.
00:07:59.680 --> 00:08:15.360
So typically that's presented to you in the form of a heat map that is projected on the geometry, and there's some coloring, typically some red spots that indicate the way you currently designed your build and you're trying to send it to the machine.
00:08:15.439 --> 00:08:16.800
This will lead to a failure.
00:08:16.959 --> 00:08:19.439
And so that's the true power of the simulation.
00:08:19.600 --> 00:08:30.800
It comes in before you actually start a printing process, and right after you have an initial idea of how your build preparation, your build job looks like.
00:08:30.959 --> 00:08:40.639
This is when you run a simulation, you get feedback, it will tell you, yes, this is good, good, you get a check, and then you send it to printer, or you will see, hey, there's a couple of red spots.
00:08:40.799 --> 00:08:52.240
You need to address those by changing the design, changing the orientation, changing the support structure, changing machine process parameters, and then redo the simulation until you have a green check mark and send it to the printer.
00:08:52.320 --> 00:08:56.480
And this will truly allow you to reduce your scrap, your failure rate.
00:08:56.639 --> 00:09:12.159
And the math is fairly easy because when you have a large printer, and let's say you know, you nowadays average build volumes for specialized materials or even standard materials are starting, you know, 250 millimeters cubic.
00:09:12.240 --> 00:09:15.360
So that's on the smaller end, and everything above is just bigger.
00:09:15.519 --> 00:09:28.080
And when you have a print that fails on that scale, that's immediately a couple, if not several thousand dollars that you lost and you cannot recoup.
00:09:28.320 --> 00:09:29.919
So you have to redo it all over again.
00:09:30.080 --> 00:09:34.159
And if you invest in a software tool, this scrap rate goes down.
00:09:34.320 --> 00:09:36.000
And it's just it's easy math.
00:09:36.159 --> 00:09:48.559
If you have a facility and you know I have this many failures a year, I can invest in software, do this prediction, optimize my processes, and considerably, if not significantly, reduce your failure.
00:09:49.039 --> 00:09:51.759
So that's the whole idea of simulation, and this is where it comes in.
00:09:52.080 --> 00:09:52.480
That's great.
00:09:52.559 --> 00:09:59.679
And I I love that idea of ROI and putting the manufacturing and additive manufacturing when you think about reduction of scrap rate.
00:09:59.840 --> 00:10:03.919
I'm curious, you know, you mentioned some variables, design, support structure.
00:10:04.159 --> 00:10:14.000
With the companies you work with, what are the most common failure modes you see that companies should be aware of as they're building this case in terms of why they might need simulation software?
00:10:14.320 --> 00:10:16.240
Yeah, I think this is a great question.
00:10:16.480 --> 00:10:48.879
Um the most common issue that I see in projects, and I've processed personally over hundreds of um customer projects, but the most common issue that I see is distortion warping that occurred due to the thermal mechanical nature of the manufacturing process, so unwanted distortion, as a matter of fact, or some sort of rupture tracking that occurs during the process or after the process.
00:10:49.039 --> 00:10:57.679
And both potentially, if not very likely, result in rendering the part not suitable for service anymore.
00:10:57.840 --> 00:11:15.360
So, for example, when you know you build a rocket part, a rocket nozzle, or some topology optimized structure for lightweight applications, a lot of materials used in metal additive manufacturing will experience some type of distortion.
00:11:15.440 --> 00:11:22.080
And that's mostly due to the subsequent heating and cooling that occurs in each and every layer.
00:11:22.240 --> 00:11:27.519
Um, and it expands and contracts, and that just leads as a consequence to distortions.
00:11:27.840 --> 00:11:39.120
And the larger the machines become and the larger parts users print, these distortions become more dominant and of bigger consideration.
00:11:39.759 --> 00:11:50.240
And so the most common failure mode is really between those two, where you see people, hey, I try to print this, but it's distorted it and I cannot use it for my application anymore.
00:11:50.399 --> 00:11:51.039
What do I do?
00:11:51.200 --> 00:12:10.879
Same question with um, hey, uh this started to delaminate from my bill plate or started to delaminate from um like where the part delaminated from the support structure and then resulted in a um build machine crash where the recoder started to hit the structure and the machine just stopped.
00:12:11.120 --> 00:12:14.080
Those are the two most common ones that I see.
00:12:14.320 --> 00:12:23.440
There are for very, very high-end application, and we're talking now very high-tech aerospace and defense industry applications.
00:12:23.759 --> 00:12:31.759
There are concerns regarding material properties and microstructure resulting due to the printing process.
00:12:31.919 --> 00:12:39.200
So this becomes of importance if you have high-end application with a lot of dynamic load exposure.
00:12:39.360 --> 00:12:49.200
The material needs to withstand um a lot of alternating loads or or high impact loads, sometimes under extremely high temperatures.
00:12:49.360 --> 00:13:05.440
And when you have an environment that is so harsh and where your printed part needs to be able to withstand all these types of loads, microstructure becomes of importance because it essentially defines how your part will respond to external loading.
00:13:05.600 --> 00:13:18.000
And so for very high-end application, it's not uncommon that users look into what microstructure results as a consequence of our printing process.
00:13:18.159 --> 00:13:37.759
So, to give you a more tangible example, when you print a large rocket nozzle, you start printing, and due to the fact in laser powder bed fusion, for example, uh, when you keep printing, the fur the further up you are in your printing process, temperature will start to accumulate in the sections that are already printed.
00:13:37.840 --> 00:13:54.879
And it's just because you keep putting heat with the laser, uh you keep putting in heat from the top, and that heat starts to accumulate, which as a consequence changes on how your melt pool, your melting operation uh per layer occurs.
00:13:55.679 --> 00:14:08.480
And to be very practical about it, or in a very simple explanation, the higher your baseline temperatures are with the same laser um parameters, your melt pull tends to become larger.
00:14:08.639 --> 00:14:21.279
That means the larger melt pool changes how the material starts to um cool down locally, and that as a consequence changes the microstructure.
00:14:21.440 --> 00:14:34.960
And this is very, very important to know, especially when you want to ensure that you need fatigue, lifetime capability of a jet engine part of a component, and that's just very important that you can investigate that.
00:14:35.200 --> 00:14:48.639
So, long story short, there are the more common problems of distortions and ruptures, and then there are some very um dedicated resource questions where simulation comes in and can help you answer those questions.
00:14:48.960 --> 00:15:00.960
I think that's such an important point, this idea of a reactive versus proactive additive workflow, because to your point, especially in those high-end specialized applications, the further downstream you get, the more expensive failure is.
00:15:01.120 --> 00:15:02.159
I mean yes, absolutely.
00:15:02.559 --> 00:15:03.440
Free print, great.
00:15:03.679 --> 00:15:05.039
Post print, painful.
00:15:05.279 --> 00:15:07.039
In the field, catastrophic.
00:15:07.200 --> 00:15:21.120
So I really like uh the way you're walking people through how this is not only an ROI on the printing portion, but it's really a huge enabler of success in the field and a good customer experience and safety.
00:15:21.279 --> 00:15:25.039
Um, I want to also talk a little bit about expertise.
00:15:25.279 --> 00:15:38.799
You know, obviously you have uh a very robust background, you've done hundreds of these yourself, you have a lot of education and experience around this, and historically, simulation has required that kind of highly specialized expertise.
00:15:39.039 --> 00:15:52.559
Help maybe the our listeners understand why has interpreting results been, I'm gonna use the word difficult, but maybe a better use of the term is required a level of expertise that someone like you has that is not widely accessible.
00:15:52.879 --> 00:15:55.759
Um yeah, that's a very good question.
00:15:56.639 --> 00:16:26.000
Um historically, when we look at how simulation, and I'm talking specifically about um finite element-based simulation, that has been typically carried out by experts that you know have gone some graduate studies through through some graduate studies and um have had years of training on how to prepare simulation models and how to interpret the results.
00:16:26.240 --> 00:16:52.960
Not specifically for additive manufacturing, but just generally speaking, the traditional career path of a simulation engineer, whether they're looking into traditional mechanical stress analyses or uh lifetime or durability predictions based on finite element results, that traditionally is carried out and or has been carried out by engineers with exhaustive training and a lot of hands-on experience.
00:16:53.120 --> 00:17:23.920
Now, when the first simulation tools, predictive simulation tools, started to appear in additive manufacturing, more specifically in metal additive manufacturing, and we're talking about um, I think the early really commercial tools were available anywhere from 2012, where you had the first solutions, till you know, really 2015 till 2018, when more and more tools started to boom and come out.
00:17:24.160 --> 00:17:33.680
But the first ones, uh the first tools based on were developed based on general finite element applications on finite element tools that were on the market.
00:17:33.839 --> 00:17:53.519
And so traditionally they had inherited the workflow, how finite element engineers traditionally used to work with these kind of tools, meaning a lot of, you know, you you need to know how to define boundary conditions, how you define material models, how you define details about running the simulation.
00:17:53.599 --> 00:17:55.519
So rather complicated a process.
00:17:55.680 --> 00:18:09.920
But then again, in 2015, around 2015, tools started to pop in, pop up that made this change a little bit easier or made prepare preparation and execution of these uh predictive runs simpler.
00:18:10.559 --> 00:18:24.319
And that I think started to democratize the simulation tools a little bit in terms of how do I uh start from a 3D geometry that I want to send to the printer to actually producing results.
00:18:24.480 --> 00:18:27.119
So that step became more democratized.
00:18:27.279 --> 00:18:34.079
But then on the latter end, as now you have simulation results, the big question, the big question is what do you do with it?
00:18:34.240 --> 00:18:48.640
Um and there, and up to this day, I think it's still a bit of a challenge because when the simulation tells you, for example, that there will be a distortion, what do you do from here?
00:18:48.880 --> 00:18:58.880
The first thing, and I'm speaking of experience with customer projects that I have been working on, is you need to understand where does this distortion, does this placement come from?
00:18:59.039 --> 00:19:01.920
Is this a buckling related issue of a thin wall?
00:19:02.079 --> 00:19:15.599
Is this something that's just driven by mechanical stresses or other features that are just pulling on the structure, why this moves out, or is this related to any other issue on the machine, why this distortion occurs?
00:19:15.839 --> 00:19:30.160
And to be honest with you, traditional simulation tools don't always provide this comprehensive picture and give you the straight answer of hey, this is a buckling issue, and that's your that's your solution to fixing it.
00:19:30.319 --> 00:19:45.759
So the reason why this is still somewhat complicated, you can identify where the issue is, what it is, but you still need some experience in terms of uh well how do I categorize this failure mechanism?
00:19:46.079 --> 00:19:48.319
How does this distortion really originate?
00:19:48.480 --> 00:19:58.960
And when once you find this out, then you have a path to a solution because depending on what type of or what causes a distortion, there's different ways to address this.
00:19:59.119 --> 00:20:06.400
To give you a very concrete example, uh a possible solution is just changing a thickness of a wall.
00:20:06.559 --> 00:20:14.640
A wall, a very thin wall that starts to buckle, you can add a thicker wall that immediately will address to a certain extent the buckling.
00:20:14.720 --> 00:20:17.119
It may not eliminate that, but it will address it.
00:20:17.279 --> 00:20:21.200
Another way to address buckling are adding support structures.
00:20:21.519 --> 00:20:32.960
Um and then nowadays there's this very, very commonly applied strategy which is referred to as counter deformation, which works really, really well.
00:20:33.119 --> 00:20:55.839
I want to say for a majority of cases, where you simulate the distortions beforehand, and then in very, very simple terms, I mean it's not that easy, but in very simple terms, just to give you an idea how this works, is you negatively scale or invert the simulated displacements, and then you send the distorted part onto the printer.
00:20:55.920 --> 00:21:02.640
And due to the tendency to distort while it's in the printer, when you take it out, it actually assumes its intended shape.
00:21:02.720 --> 00:21:10.559
So it's a beautiful uh technique that is very commonly applied in in additive manufacturing simulation that can address a lot of problems.
00:21:10.720 --> 00:21:17.279
But then you have those issues where let's say the thin walls with bucklings that you cannot address with counter deformation.
00:21:17.519 --> 00:21:22.319
And there you need still the expert knowledge to identify, hey, this is a buckling problem.
00:21:22.480 --> 00:21:25.759
I will probably not be able to address this with counter deformation.
00:21:25.920 --> 00:21:27.359
So, where do I go from here?
00:21:27.519 --> 00:21:44.160
And this is where I'm really excited about and the tools that we're developing here at Nodivac, is that AI has the capability to aid in this investigation of okay, I have a simulation results that tells me there's a distortion.
00:21:44.960 --> 00:22:03.759
And AI, if you feed it the correct information and prime it for a conversation for metal additive manufacturing, this is where it can be very useful in figuring out where does it originate, what's The sort of category, why it happened, and what can you do to fix it?
00:22:04.000 --> 00:22:08.480
So I am so glad you brought that up because that was something I wanted to explore with you as well.
00:22:08.640 --> 00:22:16.400
You have, you know, I'm going to call it a uniquely human bottleneck right now, where yes, I can run the simulation, I can have results, but so what?
00:22:16.559 --> 00:22:20.960
Tends to be still a human answer in terms of how do I mitigate or fix those issues.
00:22:21.200 --> 00:22:28.480
How are you feeding or how should people think about AI in terms of getting that human expertise enough of that in there?
00:22:28.640 --> 00:22:35.519
That then you can leverage all the benefits of AI combined with the benefits you're talking about, which are uniquely human and related to experience.
00:22:35.759 --> 00:22:40.720
How are you using those together to accelerate the use of this tool in this space?
00:22:40.960 --> 00:22:49.440
Um Yeah, I can give you a global overview on how AI is applied in the field that we are operating in.
00:22:49.599 --> 00:22:57.359
So there are groups and companies that pursue the use of AI to actually replace the simulation part.
00:22:57.519 --> 00:23:11.759
Simulations currently with the finite element method are based on mathematical relationships that are being described in the background, and you computer solve those um equations to determine a certain outcome, let's say displacements.
00:23:12.079 --> 00:23:14.720
So that's based on physics models.
00:23:14.799 --> 00:23:21.599
Um you translate into equations, numerical applications, and then you simulate that and you get a result back on your screen.
00:23:22.720 --> 00:23:38.640
One way to utilize the eye is to replace this physics-based approach with models that are either trained to produce the physics or a combination of both.
00:23:38.799 --> 00:24:01.039
So, for example, a phrase that some may be more familiar with is um physics trains neural networks, um, where you can use the true capability of AI to have the prediction of distortional stresses, um, potentially failures directly done on the actual problem.
00:24:01.200 --> 00:24:02.240
That's one part of it.
00:24:02.400 --> 00:24:04.559
And it's a very interesting field.
00:24:04.720 --> 00:24:05.200
Why?
00:24:05.519 --> 00:24:14.240
Because um simulations, uh, when you do it traditionally, like we do with uh the finite element method, they take some time.
00:24:14.400 --> 00:24:20.400
And I'm talking about now that computers have become so fast and computational power is really affordable these days.
00:24:20.559 --> 00:24:28.079
We talk in minutes and for very, very large models, maybe we talk in a couple of hours where you have to run this physics-based finite element simulation in the background.
00:24:28.240 --> 00:24:39.359
If you have a well-trained AI model, a physics-based model, the um computational time instead of hours can be reduced to sometimes seconds.
00:24:39.440 --> 00:24:43.680
If and it can be very accurate if it's very, very well trained.
00:24:43.839 --> 00:24:56.880
I think right now, where this type of technology is, it's still, I don't want to say in its infancy, but it's I don't think it has full industrial maturity because it truly matters how you train the models.
00:24:57.039 --> 00:25:10.240
And as soon as your actual case deviates from what your uh AI model was trained on, the results won't are very likely not gonna be precisely what's gonna happen in reality.
00:25:10.400 --> 00:25:21.920
So that's one of the downsides, but I think it's a super interesting field that we certainly keep our eye on it because the benefit of having a prediction almost instantaneously is just a great idea.
00:25:22.000 --> 00:25:32.559
And you know, everything that's fast is preferable, well, fast and accurate, obviously, is preferable to something that takes several minutes, maybe an hour, to come up with a prediction, right?
00:25:32.799 --> 00:25:51.119
Now, the way we bring AI into all of this is actually what you just mentioned, uh, and that is um we we try not to replace the physics part, but rather provide a what we like to call a co-engineer.
00:25:51.200 --> 00:26:07.039
So it's a trademark product that we have that is basically your simulation expert in-house that helps you investigate the results and helps you make really precise recommendations on how you can address this issue.
00:26:07.279 --> 00:26:09.359
And how did we even figure this out?
00:26:09.519 --> 00:26:10.240
This was needed.
00:26:10.480 --> 00:26:31.039
So, like I told you, I've done a lot I've done lots of customer projects, and it started to become obvious that I became the bottleneck or colleagues of mine that had an expertise on a specific topic and the customer needed to chat to them, and so they had to become available.
00:26:31.440 --> 00:26:36.160
And not only are you know it was an availability issue, but also it's costly.
00:26:36.240 --> 00:26:53.440
You know, if you have to pay an engineer that gives you expertise, feedback, they need to sit down, need to understand your case, they need to look at the simulation, and you know, you have to pay those engineers typically with a PhD or a master's degree by the hour, and it's just very costly.
00:26:53.680 --> 00:27:00.000
So when we looked at this, we started thinking, how can we get rid of this bottleneck and how can we scale this?
00:27:00.160 --> 00:27:12.480
And for a while we've been looking into generally the question, what does AI mean for us and how can we utilize it for us and provide it to our customers to help them with their cases?
00:27:12.640 --> 00:27:18.720
And that once we started sitting down and kind of reviewing this, it became obvious, hey, there's this real bottleneck, you know.
00:27:18.880 --> 00:27:22.720
Every time somebody calls us, we need to arrange a meeting and then we need to look in the case.
00:27:22.799 --> 00:27:24.480
It's just expensive, time consuming.
00:27:24.640 --> 00:27:26.880
Let's see if AI can do this for you.
00:27:26.960 --> 00:28:00.079
And when we started to develop a um prototype of linking the AI to the simulation, giving it access, priming the AI conversation for the very specific case you're working on in our tool, it became obvious this is a great solution and it it is really of great use and helps you to understand your case and gives you recommendations that I, as a you know, PhD engineer with over a decade uh of experience in additive manufacturing simulation would give you.
00:28:00.240 --> 00:28:11.039
So when we discovered this, we thought, okay, this is just a really, really great tool that can help um users the same way it helped us with our customer cases.
00:28:11.200 --> 00:28:13.519
So that's our view on AI.
00:28:13.680 --> 00:28:19.920
Keep in mind there's different ways and different um applications how AI can be applied.
00:28:20.160 --> 00:28:39.599
The one we're focusing on is not doing the simulation, but helping engineers, helping companies understand what the simulation means and how you need to change your designs to make sure the results showing from the simulations are being addressed in your build preparation before you send it to the machine.
00:28:40.000 --> 00:28:49.680
Love that idea of a co-engineer because as I'm thinking about you have the simulation software and now you are you're interjecting AI and you're getting all these efficiencies in that whole workflow.
00:28:49.839 --> 00:28:53.279
Where do you think the human expertise still matters the most?
00:28:53.599 --> 00:28:55.039
Oh yeah, it's interesting.
00:28:55.200 --> 00:29:04.000
Um, I can tell you simulation, and again, I'm a strong advocate for simulation, but it's still met with quite a bit of resistance.
00:29:04.240 --> 00:29:14.240
Um often by users that have a lot of hands-on experience, they still tend to be a little skeptical.
00:29:14.400 --> 00:29:35.759
But in all fairness, I see why they're sometimes skeptical towards simulation because they they bring also several decades of experience to the table, and their take on this is um, hey, I don't need to do simulation, I have a gut feeling on how this is gonna turn out because I've printed similar parts and I don't really need this.
00:29:35.920 --> 00:29:58.400
And to answer your question, where the human component comes in, in my opinion, it is actually really perfect for these types of engineers that already have years of experience because when they do get this gut feeling, hey, I'm not sure 100% if this is going to work, I'm gonna invest five minutes to run a simulation just to make sure that my gut feeling is right.
00:29:58.559 --> 00:30:17.920
Obviously, there's the other group as well that you know believes the simulation and and wants to do it no matter what, whether it's a beginner or a seasoned expert, but the human component where you use your best judgment, your experience for parts that you have already printed and failures that you have already experienced.
00:30:18.160 --> 00:30:23.680
That is something I don't see AI replacing soon, and that's not what we're intending.
00:30:23.839 --> 00:30:37.920
So, what we're intending is really that you have a colleague that you have a conversation with in our chat window, and they help you to understand what the numerics, what the results means from a you know, simulation and physics point of view.
00:30:38.240 --> 00:30:46.559
And that should be an additional tool to kind of comfort you and strengthen what you thought was going to happen in the first place.
00:30:46.880 --> 00:30:48.160
Yeah, I I love that.
00:30:48.319 --> 00:30:51.200
Um, and I think we we share a very similar philosophy.
00:30:51.279 --> 00:31:00.400
I mean, whether you're talking about additive manufacturing or AI or software simulation, it's like how do the people use their time for the best and highest value?
00:31:00.640 --> 00:31:01.680
That's really all it is.
00:31:01.839 --> 00:31:11.039
There's no replacement, it's being able to put the highest value on that, which is a very common lean manufacturing print uh principle, you know, in the in this idea of waste and value.
00:31:11.200 --> 00:31:12.559
And so I can really appreciate that.
00:31:12.720 --> 00:31:28.480
With since you've worked with so many customers and so many industries, you know, which maybe industries or types of customers would you say are ahead of the curve on simulation, and which industries or types of customers do you think need it most that are not ahead of the curve with simulation?
00:31:28.880 --> 00:31:35.359
Yeah, I love this question because uh it kind of shows um yeah, yeah, it's the right question to ask.
00:31:35.519 --> 00:31:40.319
So the ones that are really ahead of the curves are the bigger companies with larger budgets.
00:31:40.400 --> 00:31:46.799
So everybody at the moment that builds larger rocket aerospace parts has to certify parts.
00:31:46.960 --> 00:31:49.839
Um, companies in the automotive industry.
00:31:49.920 --> 00:31:52.640
And I mean, I don't need to drop any names here.
00:31:52.720 --> 00:31:56.079
A simple Google search of people in the industry will know the big players.
00:31:56.319 --> 00:32:05.200
For them, it's easier to allocate budget to buy traditionally very expensive simulation software for metal additive manufacturing.
00:32:05.599 --> 00:32:27.200
And due to the nature of larger companies already or having in in a lot of cases dedicated simulation teams, they can invest in research and typically process cases that are, I want to say, two to three years ahead of the rest, you know, of average players that try to produce standard parts.
00:32:27.279 --> 00:32:42.559
Um, and that involves you know applying simulation top uh methodologies to investigate very specialized materials that are attractive to the requirements of uh aerospace application or defense applications or even medical applications.
00:32:42.720 --> 00:32:55.519
So that investment shows in terms of utilizing simulation and of course other research tools and experiments to understand how we can make the 3D printing process better, how can we make better parts?
00:32:55.680 --> 00:33:01.920
I want to say they're typically a couple years ahead of everybody else, and they're um utilizing simulation.
00:33:02.000 --> 00:33:05.519
And for them, I never need to explain to them why they need simulation.
00:33:05.680 --> 00:33:10.000
They come to me with questions and they know, so that they use it, they know why they want to use it.
00:33:10.079 --> 00:33:13.759
It's more a question of hey, do you think this is the right modeling approach?
00:33:13.920 --> 00:33:16.400
Or how would you model this specific problem?
00:33:16.559 --> 00:33:18.480
So it's a different conversation with them.
00:33:18.559 --> 00:33:19.920
Um, they know their stuff.
00:33:20.160 --> 00:33:27.920
It's more like providing consultancies for modeling approaches and ping-ponging ideas of how I would do certain things.
00:33:28.079 --> 00:33:38.240
Now, the larger group um that let's say um is a service bureaus that have outputs print still print complicated parts.
00:33:38.480 --> 00:33:48.799
Um, some of them have simulation tools, but for them, simulation tools and affording simulation tools, I think is often a budgetary problem.
00:33:48.960 --> 00:33:51.519
Simulation tools are not cheap.
00:33:52.000 --> 00:34:09.920
Um, again, we are trying to change that a little bit with having very low-cost products that makes it easier for shops that have one or two machines that um still have one, two, three failures per year that you can avoid by using simulation.
00:34:10.400 --> 00:34:23.760
Um, but I feel like there's this underserved market for you know um users that have machines but not the budget to buy fifteen to you know fifty thousand dollar simulation tools.
00:34:23.920 --> 00:34:37.920
So there they typically, you know, they can probably utilize a couple of thousand dollars, maybe several hundred dollars to buy simulation tools, but not you know, twenty-thirty thousand dollars per year to to to invest in that.
00:34:38.079 --> 00:34:51.440
So those users, smaller shops, um basically working on on producing parts, they typically understand why simulation is necessary and what it can do.
00:34:51.599 --> 00:35:06.960
But practically, and again, this uh my experience also having to do a lot of sales calls with with clients that uh are interested in simulation but not quite sure whether they should invest, is there's this debate on should I truly invest.
00:35:07.199 --> 00:35:20.320
So let's say you can provide a price that is attractive to them for simulation software, there's still a lot of hidden costs that they have to be mentally prepared to swallow, and that is you need somebody to train on the software.
00:35:20.559 --> 00:35:34.079
And no matter how simple you design your predictive tool, there there will be some basic training required and some general introduction in even with our co-engineer that can help with a lot of that.
00:35:34.320 --> 00:35:37.599
You also need to show people how you know to use that.
00:35:37.760 --> 00:35:41.519
So there's this kind of mental hurdle of how do we adapt this?
00:35:41.599 --> 00:35:44.079
Well, we need to train this person on it.
00:35:44.320 --> 00:35:51.199
We have that expense, even if the chop software is very affordable, we still um need to invest in that.
00:35:51.280 --> 00:35:57.119
And then we meet we need to make sure that the person preparing the build is actually using the software.
00:35:57.280 --> 00:36:12.559
And so all these factors, in my opinion, lead to the fact that uh smaller companies and medium-sized companies are still a little hesitant because of this adoption hurdle with the aforementioned reasons.
00:36:12.800 --> 00:36:31.039
Um now I still believe that in the future this is going to change when we look at how traditional simulation, structural simulation, um grew in in adoption in automotive or in powertrain industry.
00:36:31.199 --> 00:36:37.920
It started with the bigger companies and trickled down to smaller companies using simulation tools as well.
00:36:38.079 --> 00:36:43.519
And I truly believe this will happen in metal additive manufacturing as well.
00:36:43.679 --> 00:37:03.039
And our mission is to make that happen by having a very, very attractive pricing model that works even for small companies and making it simple enough to not having to worry about training that within one or two hours of introduction you can use this and you're good to go.
00:37:03.119 --> 00:37:17.199
And then having this AI tool that helps you remembering things when you need to and helps you double-checking your um your assumptions about the simulation to get rid of this scary thing of, oh wow, how do I use simulation?
00:37:17.280 --> 00:37:18.639
Do I have to be an expert?
00:37:18.719 --> 00:37:27.199
And I think that's essentially what we're aiming for, and the benefit for for smaller companies and mid-sized companies for adapting simulation.
00:37:27.360 --> 00:37:27.599
Yeah.
00:37:27.760 --> 00:37:33.440
And I I also think the more additive manufacturing moves into cereal production.
00:37:33.519 --> 00:37:36.079
So from I mean, that's already happening now.
00:37:36.159 --> 00:37:46.159
It it I truly think there is very good cases out there where larger companies move from individual parts to what I would consider serial production.
00:37:46.320 --> 00:37:52.639
And whenever you have that, quality uh control becomes a bigger discussion.
00:37:52.880 --> 00:38:11.920
And I see simulation as well as as part of this quality control pipeline that you make sure somebody at least looked into how well can this part, uh how well can this part be printed, and are there any issues that will reveal themselves once the part goes into service.
00:38:12.239 --> 00:38:19.199
I think that's so important when you think about not only, I mean, there's so much talk in defense and elsewhere about building up our industrial base.
00:38:19.360 --> 00:38:21.920
And that is not the five Fortune 10 companies.
00:38:22.000 --> 00:38:39.360
You know, that's the swath that you're talking about of small and mid-sized manufacturers where this idea of investment isn't just in the tool, it's in the personnel, and and you have that going in parallel with what you're talking about, where we're moving into serial production and being able to address both those needs.
00:38:39.760 --> 00:38:41.360
I really applaud what y'all are doing.
00:38:41.440 --> 00:38:43.679
You know, we've covered so much ground today.
00:38:44.000 --> 00:38:53.199
If you were to approach a manufacturing leader tomorrow and they were to take one thing away uh from this whole conversation, what would you want it to be?
00:38:53.440 --> 00:39:13.039
I would encourage everybody, whether it's an OEM, a service bureau utilizing manufacturing, or really anybody that is somehow in the vicinity or neighborhood of additive manufacturing, simulation can be a wonderful tool that can really help you to make your process better.
00:39:13.199 --> 00:39:19.280
It may seem scary, it may seem that you have to overcome using another software tool in your process.
00:39:19.519 --> 00:39:32.559
But if you truly dedicate some time to understanding what it is, what it does for you on the long run, it will save you time, it will save you, as a consequence, costs.
00:39:32.639 --> 00:39:42.000
And 100% I'm convinced that it will lead to improvement of your manufacturing process and the parts that you're manufacturing.
00:39:42.159 --> 00:39:43.920
I have absolutely no doubt about that.
00:39:44.239 --> 00:39:56.159
I think that is a flawless place to conclude this conversation, and it's a great takeaway for a lot of folks that listen to this that may be on the fence in terms of can this really provide ROI?
00:39:56.239 --> 00:40:00.079
And you've laid out such a wonderful business case in addition to the technical case.
00:40:00.239 --> 00:40:06.719
So I appreciate so much you spending some time with us today and and talking about simulation software and where it can add value.
00:40:06.880 --> 00:40:08.320
It's been a wonderful conversation.
00:40:08.559 --> 00:40:09.199
Thank you very much.
00:40:09.280 --> 00:40:11.199
Uh, also kudos to you and your team.
00:40:11.360 --> 00:40:13.280
I think you're doing a great job.
00:40:13.360 --> 00:40:18.079
Um, and I I like your dedication to podcasts because I like to listen to them.
00:40:18.159 --> 00:40:19.760
That's how I discovered you guys.
00:40:19.920 --> 00:40:24.880
And I think you had great guests that I've met at trade shows that I've spoken to.
00:40:24.960 --> 00:40:28.400
Uh, and I hope you keep this up because it's just wonderful.
00:40:28.559 --> 00:40:31.599
You know, when when I ride in my car, it's along a car ride.
00:40:31.679 --> 00:40:36.400
I put on your podcast, and hey, there's this guy that I met at the trade show that year.
00:40:36.559 --> 00:40:38.320
And uh let's listen to him.
00:40:38.400 --> 00:40:39.360
What does he have to say?
00:40:39.440 --> 00:40:42.320
And then it just, you know, for myself, it generates ideas.
00:40:42.480 --> 00:40:46.639
I understand how people think about this industry that I'm very passionate about.
00:40:46.880 --> 00:40:54.159
And thank you for sharing my perspective on what I would like to at the moment still consider to be quite niche.
00:40:54.239 --> 00:40:59.519
Um, it's not, you know, simulation is not everywhere, but I'd like to think it's growing.
00:40:59.760 --> 00:41:02.239
Users more and more understand what's going on.
00:41:02.400 --> 00:41:09.360
And I hope I can I could contribute with my insight to some of the myths around simulation and how it can be adopted.
00:41:09.760 --> 00:41:11.440
Well, you absolutely did that and more.
00:41:11.760 --> 00:41:11.920
All right.
00:41:12.159 --> 00:41:19.679
One of the things that obviously was key is Christian made a compelling uh point around the business case for simulation.
00:41:20.320 --> 00:41:29.039
Tell me about you know how you think that should be fitting into the conversations that our customers or our customers that are considering additive have.
00:41:30.480 --> 00:41:37.119
I think a lot of times we think about ROI as, well, I'm printing faster, so I'm saving time, and I can do more iterations.
00:41:37.360 --> 00:41:42.239
And that's true, but sometimes a bigger ROI is preventing expensive failures.
00:41:42.400 --> 00:41:50.960
And that was really his point about leveraging simulation from a business point of view, where every downstream failure becomes exponentially more costly.
00:41:51.119 --> 00:41:56.320
And this helps those manufacturers transition from call it reactive troubleshooting, why didn't this work?
00:41:56.400 --> 00:41:59.039
What should I change to proactive manufacturing?
00:41:59.199 --> 00:42:01.360
That is the key to scale production.
00:42:01.760 --> 00:42:03.280
You know, you make such a great point.
00:42:03.360 --> 00:42:13.920
And just to, you know, when we think about manufacturing best practices, the further a par a subassembly is down the line, the more you've got invested and the more expensive a failure is.
00:42:14.079 --> 00:42:23.039
And understanding where to apply simulation and other tools, um, essentially where are the failures most costly can be a tremendous guide.
00:42:23.280 --> 00:42:27.519
You know, he also, like a lot of our guests, spent a lot of time talking about AI.
00:42:28.000 --> 00:42:33.519
What stood out to you, though, in this conversation, maybe differently from the way AI has been brought up in our other episodes?
00:42:33.920 --> 00:42:37.199
I actually loved how he referred to AI as a co-engineer.
00:42:37.440 --> 00:42:45.199
And so when he was talking about its utility, we've had a lot of conversation around AI won't replace people, you know, it'll it'll help people be more effective.
00:42:45.440 --> 00:42:53.679
But he said this accelerating expertise, rather than pretending expertise isn't needed and outsourcing that to a tool is the big difference.
00:42:53.840 --> 00:43:05.119
And if you're a co-engineer, there implies some interaction, some insight, some exchange of ideas, and that's where he sees it being best applied in his own company, but in the simulation writ large.
00:43:05.360 --> 00:43:15.119
For a long time, the bottleneck is you would take a lot of expertise to know how to set up the simulation, and then three days of a PhD to interpret the analysis.
00:43:15.360 --> 00:43:21.679
And he said, you can't replace that interpretation necessarily, but oh my gosh, could I be more effective in setting it up?
00:43:21.840 --> 00:43:25.039
And could I at least have some analysis that is done?
00:43:25.119 --> 00:43:29.599
So now I am drawing insights into so what, what do I go do with this?
00:43:29.760 --> 00:43:32.880
Not this huge litany of issues, huge accelerant.
00:43:32.960 --> 00:43:36.159
And so for forever AI, I will be referring to as my co-engineer.
00:43:36.800 --> 00:43:41.840
Well, you know, that's exciting to everyone except real engineers like me who don't the thought of a co-engineer.
00:43:41.920 --> 00:43:44.480
Now I gotta argue with somebody about what the right answer is.
00:43:45.039 --> 00:43:52.079
Um who do you think ultimately benefits the most from what Novek is building?
00:43:52.480 --> 00:43:54.079
Yes, it's a great question.
00:43:54.239 --> 00:43:59.679
I asked them something similar, but I just said, you know, who do you think is benefiting the most from from simulation pure.
00:44:00.400 --> 00:44:02.000
And who should be that isn't?
00:44:02.159 --> 00:44:09.760
And he said, like many tools, when you have a Fortune 500 or a large aerospace company, they have a lot more resources at their disposal.
00:44:09.920 --> 00:44:13.280
So they will naturally be first adopters of this type of technology.
00:44:13.519 --> 00:44:22.639
But when you think about as a percentage of my investment, who is going to reap the biggest reward or suffer the greatest consequence, it's a small and mid-sized manufacturer.
00:44:22.800 --> 00:44:26.960
Every bet they place is hugely critical to their success.
00:44:27.119 --> 00:44:30.719
And so in his view, this is who should be adopting it.
00:44:30.880 --> 00:44:46.079
And if the cost is prohibitive, meaning I can't have two full-time PhDs on staff, tools like this can help lower that barrier to entry because you're doing some of that co-engineering work via a tool where before you'd have to have it via full-time FT FTEs.
00:44:46.400 --> 00:44:50.400
You touched on something there that I think is so important and often overlooked.
00:44:50.559 --> 00:44:55.199
And that is the smaller the company, the better your batting average has to be.
00:44:55.360 --> 00:44:55.599
Yes.
00:44:55.840 --> 00:44:56.079
Right?
00:44:56.159 --> 00:45:06.800
If I am, you know, I don't know, alphabet, Toyota, pick a pick a large company, how many bets are they making every month, every quarter, every year?
00:45:07.039 --> 00:45:08.880
Hundreds and hundreds and hundreds.
00:45:09.119 --> 00:45:19.920
If you are a company, you know, that is a you know, ten million dollar, hundred million dollar, five hundred million dollar company, you're making a relatively small number of bets.
00:45:20.079 --> 00:45:25.199
And the further you get down that size scale, the more you're counting on those bets to work.
00:45:25.760 --> 00:45:44.239
And I think his point around, again, what's the importance of investing where there are a lot of small and medium-sized business owners who I think would do well to consider, you know, your strengths are agility, ability to tailor a customer experience more easily for different customer types.
00:45:44.559 --> 00:45:52.239
But how do you how do you play to those strengths and then minimize the fact that my batting average has to be better than a lot of my larger competitors?
00:45:52.400 --> 00:45:54.880
And I think it's a great tool to improve that.
00:45:57.119 --> 00:46:06.960
We spend a lot of time talking about how to go from pilots kind of prototyping with additive into full-blown production at scale.
00:46:07.280 --> 00:46:11.599
How do you think simulation fits into that equation based on what he shared?
00:46:12.000 --> 00:46:16.639
Yeah, you know, when you think about what is required, production requires predictability.
00:46:16.800 --> 00:46:20.000
And predictability requires understanding your process.
00:46:20.239 --> 00:46:26.800
And a lot of times when you think about simulation, sure, it's failure prevention, but it's really a quality control tool.
00:46:26.960 --> 00:46:29.679
It's to help ensure that you have that consistent process.
00:46:29.760 --> 00:46:33.280
Um, and it's no different than inspection or process validation.
00:46:33.360 --> 00:46:39.039
And so I think that I have confidence that I understand my process and it'll work before I even hit print is huge.
00:46:39.199 --> 00:46:50.239
And we're talking about simulation, and it's used most often in the metal world, but that same mindset we have our VP here at B9 Creations always talks about building quality into each stage.
00:46:50.400 --> 00:46:52.480
That doesn't mean once you start building.
00:46:52.639 --> 00:46:54.719
That's even in the pre-build stage.
00:46:54.800 --> 00:47:07.360
And so I really appreciated his point around predictability and process and how this is such a vital part when you're having something that is very expensive, time and money, if you don't take the time to figure out what your failure points are.
00:47:08.159 --> 00:47:08.719
Agreed.
00:47:09.039 --> 00:47:10.480
Couldn't have said it better.
00:47:10.719 --> 00:47:19.280
Um, you know, this question we've started to ask if one of our listeners only took away one thing from this episode, what do you think that should be?
00:47:19.599 --> 00:47:23.679
Yeah, it was great talking to someone who uh doesn't live in the hardware and material space.
00:47:23.840 --> 00:47:34.079
Not those aren't critical, but we've talked a lot about uh with guests about the future of additive, you know, you need better technology, better materials, but really when you boil down to it, you need a smarter workflow.
00:47:34.239 --> 00:47:35.920
And that's what I would take away from this.
00:47:36.079 --> 00:47:38.400
And his work sits at the intersection of that.
00:47:38.559 --> 00:47:51.440
And I think it's if I have a smarter workflow that reduces scrap, makes me more money, helps improve repeatability, where I love that AI combination is then how can I democratize it and make it more accessible?
00:47:51.679 --> 00:48:01.760
So smarter, more accessible workflows, applying that methodology across everywhere you are leveraging advanced manufacturing tech is probably my one takeaway from that conversation.
00:48:02.159 --> 00:48:13.119
And I think that relates back quite a bit to the small and medium-sized businesses thinking about where they should be leveraging this to address some of the natural vulnerabilities of a smaller business.
00:48:13.280 --> 00:48:19.840
Um I wanna I want to touch on something that's coming up for everybody who is you know following along at home.
00:48:20.159 --> 00:48:23.280
So many of these conversations touch on the topic of workforce.
00:48:23.840 --> 00:48:31.360
And the lady behind the curtain, Nicole, had a great idea that we ought to do a series on workforce, like we did a series on Made in America.
00:48:31.440 --> 00:48:50.639
And so we're putting together, at this point, not quite sure if it's gonna be three, four, or five parts, but talking to policymakers, educational experts, industry experts around this topic because we hear it all the time, but so many times you hear you hear a word too many times, it kind of loses its meaning.
00:48:50.800 --> 00:48:56.239
We're gonna peel that onion and and dig deep into that in a in some future episodes and stay tuned.
00:48:56.639 --> 00:48:56.960
Love it.
00:48:57.119 --> 00:49:03.360
One of the biggest takeaways from this conversation is that simulation isn't really about software, it's about confidence.
00:49:03.599 --> 00:49:13.039
And confidence that your build will succeed before you spend thousands of dollars on materials and machine time, confidence your process is repeatable and as you move towards production.
00:49:13.119 --> 00:49:23.119
And ultimately, I think there's this idea that confidence needs to happen for additive manufacturing to be a reliable process rather than a series of expensive experiments.
00:49:23.280 --> 00:49:31.760
So we thank Christian so much for joining us and sharing his technical expertise and his vision for making this simulation more accessible to manufacturers everywhere.
00:49:32.159 --> 00:49:35.760
Thanks again for joining us on this episode of the Additive Advantage Podcast.
00:49:35.920 --> 00:49:45.760
If you enjoyed this conversation, be sure to follow this show on Apple Podcasts or Spotify, subscribe on YouTube, and connect with us on LinkedIn so that you never miss a new episode.
00:49:46.000 --> 00:49:51.440
Until next time, keep building, keep learning, keep finding your additive advantage.
00:00:00.000 --> 00:00:02.879
When you do this at home with your plastic printer, not a big deal.
00:00:03.040 --> 00:00:07.440
It takes a couple of minutes, you know, to reset the printer and to start printing the process.
00:00:07.599 --> 00:00:24.800
But nowadays, when we look at metal, industrial metal additive manufacturing printers that print very large parts, some to you know, up to about a cubic meter of build volume, where they print rocket parts or very large parts for the defense industry.
00:00:25.519 --> 00:00:38.399
If a failure occurs when you print such a large part, it's a huge waste because chances are very, very slim that you can restart this printing process and continue when this failure occurred.
00:00:38.560 --> 00:00:42.000
So, more likely, what you have to do is you have to take everything out.
00:00:42.159 --> 00:00:48.880
It's basically considered scrap, and you have wasted time and money to try to build a part.
00:00:49.039 --> 00:00:50.880
Now, this is where simulation comes in.
00:00:51.119 --> 00:00:53.200
Welcome to the Additive Advantage Podcast.
00:00:53.359 --> 00:00:54.640
I'm Sean Anderson.
00:00:54.960 --> 00:01:01.520
This is where we explore people, technologies, and strategies that are shaping the future of advanced manufacturing.
00:01:01.920 --> 00:01:04.239
My co-host Danny Mason and I.
00:01:05.200 --> 00:01:06.879
Yeah, I'm sorry, I don't know what.
00:01:07.040 --> 00:01:09.040
Clearly it's not 9 30 at night.
00:01:09.120 --> 00:01:09.760
Yeah.
00:01:11.439 --> 00:01:15.359
I was reading two paragraphs at once and just kind of randomly mixing them together.
00:01:17.840 --> 00:01:19.599
That's how nope, this is all good.
00:01:19.760 --> 00:01:22.159
That is how my brain works all the time.
00:01:23.680 --> 00:01:31.599
Welcome to the Additive Advantage Podcast, where we explore the people, technologies, and strategies that are shaping the future of advanced manufacturing.
00:01:31.840 --> 00:01:34.959
I'm Sean Anderson, here with my co-host Danny Mason.
00:01:35.280 --> 00:01:45.519
Whether you are leading an additive program, evaluating technologies, or just looking for practical ways to improve manufacturing outcomes, we're glad you're joining us.
00:01:46.159 --> 00:01:56.239
Today's conversation tackles a topic that often flies under the radar in additive manufacturing, but arguably has one of the biggest impacts on production success, simulation.
00:01:56.480 --> 00:02:03.599
If you've ever had a metal print fail after hours or days of build time, you know how expensive trial and error can become.
00:02:03.760 --> 00:02:07.359
But what if you could identify those failures before ever pressing print?
00:02:07.599 --> 00:02:11.840
Joining us today is Christian Rossman, founder and CEO of Notavek.
00:02:12.080 --> 00:02:23.680
Christian has spent more than a decade at the intersection of metal additive, simulation, and predictive engineering, including time at Materialize and founding Additive Lab before launching Notavek.
00:02:24.000 --> 00:02:36.960
In this episode, we explore how simulation is evolving from a tool reserved for large companies into something accessible for manufacturers of every size, and how AI may be the catalyst that finally brings simulation into the mainstream.
00:02:37.120 --> 00:02:37.599
Let's dive in.
00:02:52.159 --> 00:02:54.080
All right, so my name is Christian Rossmann.
00:02:54.159 --> 00:02:55.199
I was born in Austria.
00:02:55.280 --> 00:02:56.800
This is where I grew up.
00:02:57.039 --> 00:03:02.479
Um and I followed an academic career for quite some time.
00:03:02.560 --> 00:03:12.479
So I studied mechanical engineering, um, did a master's in business engineering, and um got my PhD in computer science.
00:03:12.639 --> 00:03:20.159
And um that PhD was partially completed in um at the medical university here in South Carolina.
00:03:20.240 --> 00:03:26.800
So this was the first time I actually moved to the US and uh got to experience what life in the US is like.
00:03:26.960 --> 00:03:35.280
Then in 2014, my wife and I we moved to Belgium, and when we did that, I was actually looking for a job.
00:03:35.439 --> 00:03:40.800
So traditionally, my experience was in the traditional mechanical engineering field.
00:03:40.960 --> 00:03:53.120
But in 2014, when we moved back to Europe, I got a job at this company Materialize, which I'm pretty sure most of your audience uh know this company.
00:03:53.680 --> 00:04:08.639
And there I um got exposed to metal additive manufacturing, more specifically, how you can use numerical tools, predictive tools to um investigate what's going on in the printing process.
00:04:08.879 --> 00:04:11.599
And I'm assuming we will talk about this a little later.
00:04:11.759 --> 00:04:17.279
So there's a lot of things that can happen that result in failures which cost companies money.
00:04:17.519 --> 00:04:28.000
Um and so yeah, I got exposure to that, and we started investigating different processes um uh and different methods and how to do these predictions.
00:04:28.160 --> 00:04:51.519
And then in 2017, 2018 is where I found my first company, Additive Lab, um, in Belgium, that um was basically a research tool with um a variety of um numerical tools and applications to do to answer questions that one may have when it comes up to the metal additive manufacturing process.
00:04:51.600 --> 00:05:00.079
And in 2022, I my wife and I and my my family with our child, we moved back to the US.
00:05:01.199 --> 00:05:13.199
And this year, with having you know more than 15 years of experience in the additive manufacturing simulation field, specifically on the process prediction side.
00:05:14.560 --> 00:05:34.319
Um, I felt a need there's I felt there's a market need for a new tool with um a lower a lower cost burden that allows users with lesser budget to utilize the advantages of simulation and the predictive power to optimize and improve their metal additive manufacturing processes.
00:05:34.480 --> 00:05:35.120
It's wonderful.
00:05:35.279 --> 00:05:38.639
What a what a robust history too in the additive field.
00:05:38.800 --> 00:05:41.199
And you've touched so many different facets of it.
00:05:41.360 --> 00:05:50.160
I think what would be helpful, um, because our audience, some people are familiar, you know, with why you would do pre-print analysis or simulation if they have done metal additive.
00:05:50.240 --> 00:05:52.399
A lot of folks maybe have less familiarity.
00:05:52.639 --> 00:06:00.480
For someone who is unfamiliar with this process, what types of things can a simulation tell them before a part is ever printed?
00:06:00.720 --> 00:06:03.199
Yeah, so to me, I'm very passionate about it.
00:06:03.279 --> 00:06:04.959
I'm gonna be very honest with you.
00:06:05.199 --> 00:06:13.839
I do not understand quite why not everybody with a metal 3D printer is using predictive simulation software.
00:06:14.000 --> 00:06:36.000
So when you have a metal additive manufacturing process like laser powder bit fusion, direct energy deposition, um sinter-based processes like metal binder jetting or cold metal fusion, all these processes, there's a lot of complicated physics happening that can lead to manufacturing failures.
00:06:36.240 --> 00:06:44.240
So everybody, even with a with a desktop 3 printer at home, I'm sure has experience you start the printing job and then all of a sudden the job fails.
00:06:44.319 --> 00:06:46.000
So you have to redo it all over again.
00:06:46.079 --> 00:06:48.959
When you do this at home with your plastic printer, not a big deal.
00:06:49.120 --> 00:06:53.519
It takes a couple of minutes, you know, to reset the printer and to start printing the process.
00:06:53.759 --> 00:07:10.879
But nowadays, when we look at metal, industrial metal additive manufacturing printers that print very large parts, some to you know, up to about a cubic meter of build volume where they print rocket parts or very large parts for the defense industry.
00:07:11.600 --> 00:07:24.480
If a failure occurs when you print such a large part, it's a huge waste because chances are very, very slim that you can restart this printing process and continue when this failure occurred.
00:07:24.639 --> 00:07:28.079
So, more likely, what you have to do is you have to take everything out.
00:07:28.240 --> 00:07:34.959
It's basically considered scrap, and you have wasted time and money to try to build a part.
00:07:35.120 --> 00:07:37.040
Now, this is where simulation comes in.
00:07:37.120 --> 00:07:51.040
So, what simulation does is typically in form as a uh of a software tool that takes in, you know, for the most part, a couple of minutes, maybe up to an hour of time, that you load into uh your geometry.
00:07:51.199 --> 00:07:59.519
So when you prepare your build, typically some 3D geometries, you run the simulation and it will tell you if there are certain problems.
00:07:59.680 --> 00:08:15.360
So typically that's presented to you in the form of a heat map that is projected on the geometry, and there's some coloring, typically some red spots that indicate the way you currently designed your build and you're trying to send it to the machine.
00:08:15.439 --> 00:08:16.800
This will lead to a failure.
00:08:16.959 --> 00:08:19.439
And so that's the true power of the simulation.
00:08:19.600 --> 00:08:30.800
It comes in before you actually start a printing process, and right after you have an initial idea of how your build preparation, your build job looks like.
00:08:30.959 --> 00:08:40.639
This is when you run a simulation, you get feedback, it will tell you, yes, this is good, good, you get a check, and then you send it to printer, or you will see, hey, there's a couple of red spots.
00:08:40.799 --> 00:08:52.240
You need to address those by changing the design, changing the orientation, changing the support structure, changing machine process parameters, and then redo the simulation until you have a green check mark and send it to the printer.
00:08:52.320 --> 00:08:56.480
And this will truly allow you to reduce your scrap, your failure rate.
00:08:56.639 --> 00:09:12.159
And the math is fairly easy because when you have a large printer, and let's say you know, you nowadays average build volumes for specialized materials or even standard materials are starting, you know, 250 millimeters cubic.
00:09:12.240 --> 00:09:15.360
So that's on the smaller end, and everything above is just bigger.
00:09:15.519 --> 00:09:28.080
And when you have a print that fails on that scale, that's immediately a couple, if not several thousand dollars that you lost and you cannot recoup.
00:09:28.320 --> 00:09:29.919
So you have to redo it all over again.
00:09:30.080 --> 00:09:34.159
And if you invest in a software tool, this scrap rate goes down.
00:09:34.320 --> 00:09:36.000
And it's just it's easy math.
00:09:36.159 --> 00:09:48.559
If you have a facility and you know I have this many failures a year, I can invest in software, do this prediction, optimize my processes, and considerably, if not significantly, reduce your failure.
00:09:49.039 --> 00:09:51.759
So that's the whole idea of simulation, and this is where it comes in.
00:09:52.080 --> 00:09:52.480
That's great.
00:09:52.559 --> 00:09:59.679
And I I love that idea of ROI and putting the manufacturing and additive manufacturing when you think about reduction of scrap rate.
00:09:59.840 --> 00:10:03.919
I'm curious, you know, you mentioned some variables, design, support structure.
00:10:04.159 --> 00:10:14.000
With the companies you work with, what are the most common failure modes you see that companies should be aware of as they're building this case in terms of why they might need simulation software?
00:10:14.320 --> 00:10:16.240
Yeah, I think this is a great question.
00:10:16.480 --> 00:10:48.879
Um the most common issue that I see in projects, and I've processed personally over hundreds of um customer projects, but the most common issue that I see is distortion warping that occurred due to the thermal mechanical nature of the manufacturing process, so unwanted distortion, as a matter of fact, or some sort of rupture tracking that occurs during the process or after the process.
00:10:49.039 --> 00:10:57.679
And both potentially, if not very likely, result in rendering the part not suitable for service anymore.
00:10:57.840 --> 00:11:15.360
So, for example, when you know you build a rocket part, a rocket nozzle, or some topology optimized structure for lightweight applications, a lot of materials used in metal additive manufacturing will experience some type of distortion.
00:11:15.440 --> 00:11:22.080
And that's mostly due to the subsequent heating and cooling that occurs in each and every layer.
00:11:22.240 --> 00:11:27.519
Um, and it expands and contracts, and that just leads as a consequence to distortions.
00:11:27.840 --> 00:11:39.120
And the larger the machines become and the larger parts users print, these distortions become more dominant and of bigger consideration.
00:11:39.759 --> 00:11:50.240
And so the most common failure mode is really between those two, where you see people, hey, I try to print this, but it's distorted it and I cannot use it for my application anymore.
00:11:50.399 --> 00:11:51.039
What do I do?
00:11:51.200 --> 00:12:10.879
Same question with um, hey, uh this started to delaminate from my bill plate or started to delaminate from um like where the part delaminated from the support structure and then resulted in a um build machine crash where the recoder started to hit the structure and the machine just stopped.
00:12:11.120 --> 00:12:14.080
Those are the two most common ones that I see.
00:12:14.320 --> 00:12:23.440
There are for very, very high-end application, and we're talking now very high-tech aerospace and defense industry applications.
00:12:23.759 --> 00:12:31.759
There are concerns regarding material properties and microstructure resulting due to the printing process.
00:12:31.919 --> 00:12:39.200
So this becomes of importance if you have high-end application with a lot of dynamic load exposure.
00:12:39.360 --> 00:12:49.200
The material needs to withstand um a lot of alternating loads or or high impact loads, sometimes under extremely high temperatures.
00:12:49.360 --> 00:13:05.440
And when you have an environment that is so harsh and where your printed part needs to be able to withstand all these types of loads, microstructure becomes of importance because it essentially defines how your part will respond to external loading.
00:13:05.600 --> 00:13:18.000
And so for very high-end application, it's not uncommon that users look into what microstructure results as a consequence of our printing process.
00:13:18.159 --> 00:13:37.759
So, to give you a more tangible example, when you print a large rocket nozzle, you start printing, and due to the fact in laser powder bed fusion, for example, uh, when you keep printing, the fur the further up you are in your printing process, temperature will start to accumulate in the sections that are already printed.
00:13:37.840 --> 00:13:54.879
And it's just because you keep putting heat with the laser, uh you keep putting in heat from the top, and that heat starts to accumulate, which as a consequence changes on how your melt pool, your melting operation uh per layer occurs.
00:13:55.679 --> 00:14:08.480
And to be very practical about it, or in a very simple explanation, the higher your baseline temperatures are with the same laser um parameters, your melt pull tends to become larger.
00:14:08.639 --> 00:14:21.279
That means the larger melt pool changes how the material starts to um cool down locally, and that as a consequence changes the microstructure.
00:14:21.440 --> 00:14:34.960
And this is very, very important to know, especially when you want to ensure that you need fatigue, lifetime capability of a jet engine part of a component, and that's just very important that you can investigate that.
00:14:35.200 --> 00:14:48.639
So, long story short, there are the more common problems of distortions and ruptures, and then there are some very um dedicated resource questions where simulation comes in and can help you answer those questions.
00:14:48.960 --> 00:15:00.960
I think that's such an important point, this idea of a reactive versus proactive additive workflow, because to your point, especially in those high-end specialized applications, the further downstream you get, the more expensive failure is.
00:15:01.120 --> 00:15:02.159
I mean yes, absolutely.
00:15:02.559 --> 00:15:03.440
Free print, great.
00:15:03.679 --> 00:15:05.039
Post print, painful.
00:15:05.279 --> 00:15:07.039
In the field, catastrophic.
00:15:07.200 --> 00:15:21.120
So I really like uh the way you're walking people through how this is not only an ROI on the printing portion, but it's really a huge enabler of success in the field and a good customer experience and safety.
00:15:21.279 --> 00:15:25.039
Um, I want to also talk a little bit about expertise.
00:15:25.279 --> 00:15:38.799
You know, obviously you have uh a very robust background, you've done hundreds of these yourself, you have a lot of education and experience around this, and historically, simulation has required that kind of highly specialized expertise.
00:15:39.039 --> 00:15:52.559
Help maybe the our listeners understand why has interpreting results been, I'm gonna use the word difficult, but maybe a better use of the term is required a level of expertise that someone like you has that is not widely accessible.
00:15:52.879 --> 00:15:55.759
Um yeah, that's a very good question.
00:15:56.639 --> 00:16:26.000
Um historically, when we look at how simulation, and I'm talking specifically about um finite element-based simulation, that has been typically carried out by experts that you know have gone some graduate studies through through some graduate studies and um have had years of training on how to prepare simulation models and how to interpret the results.
00:16:26.240 --> 00:16:52.960
Not specifically for additive manufacturing, but just generally speaking, the traditional career path of a simulation engineer, whether they're looking into traditional mechanical stress analyses or uh lifetime or durability predictions based on finite element results, that traditionally is carried out and or has been carried out by engineers with exhaustive training and a lot of hands-on experience.
00:16:53.120 --> 00:17:23.920
Now, when the first simulation tools, predictive simulation tools, started to appear in additive manufacturing, more specifically in metal additive manufacturing, and we're talking about um, I think the early really commercial tools were available anywhere from 2012, where you had the first solutions, till you know, really 2015 till 2018, when more and more tools started to boom and come out.
00:17:24.160 --> 00:17:33.680
But the first ones, uh the first tools based on were developed based on general finite element applications on finite element tools that were on the market.
00:17:33.839 --> 00:17:53.519
And so traditionally they had inherited the workflow, how finite element engineers traditionally used to work with these kind of tools, meaning a lot of, you know, you you need to know how to define boundary conditions, how you define material models, how you define details about running the simulation.
00:17:53.599 --> 00:17:55.519
So rather complicated a process.
00:17:55.680 --> 00:18:09.920
But then again, in 2015, around 2015, tools started to pop in, pop up that made this change a little bit easier or made prepare preparation and execution of these uh predictive runs simpler.
00:18:10.559 --> 00:18:24.319
And that I think started to democratize the simulation tools a little bit in terms of how do I uh start from a 3D geometry that I want to send to the printer to actually producing results.
00:18:24.480 --> 00:18:27.119
So that step became more democratized.
00:18:27.279 --> 00:18:34.079
But then on the latter end, as now you have simulation results, the big question, the big question is what do you do with it?
00:18:34.240 --> 00:18:48.640
Um and there, and up to this day, I think it's still a bit of a challenge because when the simulation tells you, for example, that there will be a distortion, what do you do from here?
00:18:48.880 --> 00:18:58.880
The first thing, and I'm speaking of experience with customer projects that I have been working on, is you need to understand where does this distortion, does this placement come from?
00:18:59.039 --> 00:19:01.920
Is this a buckling related issue of a thin wall?
00:19:02.079 --> 00:19:15.599
Is this something that's just driven by mechanical stresses or other features that are just pulling on the structure, why this moves out, or is this related to any other issue on the machine, why this distortion occurs?
00:19:15.839 --> 00:19:30.160
And to be honest with you, traditional simulation tools don't always provide this comprehensive picture and give you the straight answer of hey, this is a buckling issue, and that's your that's your solution to fixing it.
00:19:30.319 --> 00:19:45.759
So the reason why this is still somewhat complicated, you can identify where the issue is, what it is, but you still need some experience in terms of uh well how do I categorize this failure mechanism?
00:19:46.079 --> 00:19:48.319
How does this distortion really originate?
00:19:48.480 --> 00:19:58.960
And when once you find this out, then you have a path to a solution because depending on what type of or what causes a distortion, there's different ways to address this.
00:19:59.119 --> 00:20:06.400
To give you a very concrete example, uh a possible solution is just changing a thickness of a wall.
00:20:06.559 --> 00:20:14.640
A wall, a very thin wall that starts to buckle, you can add a thicker wall that immediately will address to a certain extent the buckling.
00:20:14.720 --> 00:20:17.119
It may not eliminate that, but it will address it.
00:20:17.279 --> 00:20:21.200
Another way to address buckling are adding support structures.
00:20:21.519 --> 00:20:32.960
Um and then nowadays there's this very, very commonly applied strategy which is referred to as counter deformation, which works really, really well.
00:20:33.119 --> 00:20:55.839
I want to say for a majority of cases, where you simulate the distortions beforehand, and then in very, very simple terms, I mean it's not that easy, but in very simple terms, just to give you an idea how this works, is you negatively scale or invert the simulated displacements, and then you send the distorted part onto the printer.
00:20:55.920 --> 00:21:02.640
And due to the tendency to distort while it's in the printer, when you take it out, it actually assumes its intended shape.
00:21:02.720 --> 00:21:10.559
So it's a beautiful uh technique that is very commonly applied in in additive manufacturing simulation that can address a lot of problems.
00:21:10.720 --> 00:21:17.279
But then you have those issues where let's say the thin walls with bucklings that you cannot address with counter deformation.
00:21:17.519 --> 00:21:22.319
And there you need still the expert knowledge to identify, hey, this is a buckling problem.
00:21:22.480 --> 00:21:25.759
I will probably not be able to address this with counter deformation.
00:21:25.920 --> 00:21:27.359
So, where do I go from here?
00:21:27.519 --> 00:21:44.160
And this is where I'm really excited about and the tools that we're developing here at Nodivac, is that AI has the capability to aid in this investigation of okay, I have a simulation results that tells me there's a distortion.
00:21:44.960 --> 00:22:03.759
And AI, if you feed it the correct information and prime it for a conversation for metal additive manufacturing, this is where it can be very useful in figuring out where does it originate, what's The sort of category, why it happened, and what can you do to fix it?
00:22:04.000 --> 00:22:08.480
So I am so glad you brought that up because that was something I wanted to explore with you as well.
00:22:08.640 --> 00:22:16.400
You have, you know, I'm going to call it a uniquely human bottleneck right now, where yes, I can run the simulation, I can have results, but so what?
00:22:16.559 --> 00:22:20.960
Tends to be still a human answer in terms of how do I mitigate or fix those issues.
00:22:21.200 --> 00:22:28.480
How are you feeding or how should people think about AI in terms of getting that human expertise enough of that in there?
00:22:28.640 --> 00:22:35.519
That then you can leverage all the benefits of AI combined with the benefits you're talking about, which are uniquely human and related to experience.
00:22:35.759 --> 00:22:40.720
How are you using those together to accelerate the use of this tool in this space?
00:22:40.960 --> 00:22:49.440
Um Yeah, I can give you a global overview on how AI is applied in the field that we are operating in.
00:22:49.599 --> 00:22:57.359
So there are groups and companies that pursue the use of AI to actually replace the simulation part.
00:22:57.519 --> 00:23:11.759
Simulations currently with the finite element method are based on mathematical relationships that are being described in the background, and you computer solve those um equations to determine a certain outcome, let's say displacements.
00:23:12.079 --> 00:23:14.720
So that's based on physics models.
00:23:14.799 --> 00:23:21.599
Um you translate into equations, numerical applications, and then you simulate that and you get a result back on your screen.
00:23:22.720 --> 00:23:38.640
One way to utilize the eye is to replace this physics-based approach with models that are either trained to produce the physics or a combination of both.
00:23:38.799 --> 00:24:01.039
So, for example, a phrase that some may be more familiar with is um physics trains neural networks, um, where you can use the true capability of AI to have the prediction of distortional stresses, um, potentially failures directly done on the actual problem.
00:24:01.200 --> 00:24:02.240
That's one part of it.
00:24:02.400 --> 00:24:04.559
And it's a very interesting field.
00:24:04.720 --> 00:24:05.200
Why?
00:24:05.519 --> 00:24:14.240
Because um simulations, uh, when you do it traditionally, like we do with uh the finite element method, they take some time.
00:24:14.400 --> 00:24:20.400
And I'm talking about now that computers have become so fast and computational power is really affordable these days.
00:24:20.559 --> 00:24:28.079
We talk in minutes and for very, very large models, maybe we talk in a couple of hours where you have to run this physics-based finite element simulation in the background.
00:24:28.240 --> 00:24:39.359
If you have a well-trained AI model, a physics-based model, the um computational time instead of hours can be reduced to sometimes seconds.
00:24:39.440 --> 00:24:43.680
If and it can be very accurate if it's very, very well trained.
00:24:43.839 --> 00:24:56.880
I think right now, where this type of technology is, it's still, I don't want to say in its infancy, but it's I don't think it has full industrial maturity because it truly matters how you train the models.
00:24:57.039 --> 00:25:10.240
And as soon as your actual case deviates from what your uh AI model was trained on, the results won't are very likely not gonna be precisely what's gonna happen in reality.
00:25:10.400 --> 00:25:21.920
So that's one of the downsides, but I think it's a super interesting field that we certainly keep our eye on it because the benefit of having a prediction almost instantaneously is just a great idea.
00:25:22.000 --> 00:25:32.559
And you know, everything that's fast is preferable, well, fast and accurate, obviously, is preferable to something that takes several minutes, maybe an hour, to come up with a prediction, right?
00:25:32.799 --> 00:25:51.119
Now, the way we bring AI into all of this is actually what you just mentioned, uh, and that is um we we try not to replace the physics part, but rather provide a what we like to call a co-engineer.
00:25:51.200 --> 00:26:07.039
So it's a trademark product that we have that is basically your simulation expert in-house that helps you investigate the results and helps you make really precise recommendations on how you can address this issue.
00:26:07.279 --> 00:26:09.359
And how did we even figure this out?
00:26:09.519 --> 00:26:10.240
This was needed.
00:26:10.480 --> 00:26:31.039
So, like I told you, I've done a lot I've done lots of customer projects, and it started to become obvious that I became the bottleneck or colleagues of mine that had an expertise on a specific topic and the customer needed to chat to them, and so they had to become available.
00:26:31.440 --> 00:26:36.160
And not only are you know it was an availability issue, but also it's costly.
00:26:36.240 --> 00:26:53.440
You know, if you have to pay an engineer that gives you expertise, feedback, they need to sit down, need to understand your case, they need to look at the simulation, and you know, you have to pay those engineers typically with a PhD or a master's degree by the hour, and it's just very costly.
00:26:53.680 --> 00:27:00.000
So when we looked at this, we started thinking, how can we get rid of this bottleneck and how can we scale this?
00:27:00.160 --> 00:27:12.480
And for a while we've been looking into generally the question, what does AI mean for us and how can we utilize it for us and provide it to our customers to help them with their cases?
00:27:12.640 --> 00:27:18.720
And that once we started sitting down and kind of reviewing this, it became obvious, hey, there's this real bottleneck, you know.
00:27:18.880 --> 00:27:22.720
Every time somebody calls us, we need to arrange a meeting and then we need to look in the case.
00:27:22.799 --> 00:27:24.480
It's just expensive, time consuming.
00:27:24.640 --> 00:27:26.880
Let's see if AI can do this for you.
00:27:26.960 --> 00:28:00.079
And when we started to develop a um prototype of linking the AI to the simulation, giving it access, priming the AI conversation for the very specific case you're working on in our tool, it became obvious this is a great solution and it it is really of great use and helps you to understand your case and gives you recommendations that I, as a you know, PhD engineer with over a decade uh of experience in additive manufacturing simulation would give you.
00:28:00.240 --> 00:28:11.039
So when we discovered this, we thought, okay, this is just a really, really great tool that can help um users the same way it helped us with our customer cases.
00:28:11.200 --> 00:28:13.519
So that's our view on AI.
00:28:13.680 --> 00:28:19.920
Keep in mind there's different ways and different um applications how AI can be applied.
00:28:20.160 --> 00:28:39.599
The one we're focusing on is not doing the simulation, but helping engineers, helping companies understand what the simulation means and how you need to change your designs to make sure the results showing from the simulations are being addressed in your build preparation before you send it to the machine.
00:28:40.000 --> 00:28:49.680
Love that idea of a co-engineer because as I'm thinking about you have the simulation software and now you are you're interjecting AI and you're getting all these efficiencies in that whole workflow.
00:28:49.839 --> 00:28:53.279
Where do you think the human expertise still matters the most?
00:28:53.599 --> 00:28:55.039
Oh yeah, it's interesting.
00:28:55.200 --> 00:29:04.000
Um, I can tell you simulation, and again, I'm a strong advocate for simulation, but it's still met with quite a bit of resistance.
00:29:04.240 --> 00:29:14.240
Um often by users that have a lot of hands-on experience, they still tend to be a little skeptical.
00:29:14.400 --> 00:29:35.759
But in all fairness, I see why they're sometimes skeptical towards simulation because they they bring also several decades of experience to the table, and their take on this is um, hey, I don't need to do simulation, I have a gut feeling on how this is gonna turn out because I've printed similar parts and I don't really need this.
00:29:35.920 --> 00:29:58.400
And to answer your question, where the human component comes in, in my opinion, it is actually really perfect for these types of engineers that already have years of experience because when they do get this gut feeling, hey, I'm not sure 100% if this is going to work, I'm gonna invest five minutes to run a simulation just to make sure that my gut feeling is right.
00:29:58.559 --> 00:30:17.920
Obviously, there's the other group as well that you know believes the simulation and and wants to do it no matter what, whether it's a beginner or a seasoned expert, but the human component where you use your best judgment, your experience for parts that you have already printed and failures that you have already experienced.
00:30:18.160 --> 00:30:23.680
That is something I don't see AI replacing soon, and that's not what we're intending.
00:30:23.839 --> 00:30:37.920
So, what we're intending is really that you have a colleague that you have a conversation with in our chat window, and they help you to understand what the numerics, what the results means from a you know, simulation and physics point of view.
00:30:38.240 --> 00:30:46.559
And that should be an additional tool to kind of comfort you and strengthen what you thought was going to happen in the first place.
00:30:46.880 --> 00:30:48.160
Yeah, I I love that.
00:30:48.319 --> 00:30:51.200
Um, and I think we we share a very similar philosophy.
00:30:51.279 --> 00:31:00.400
I mean, whether you're talking about additive manufacturing or AI or software simulation, it's like how do the people use their time for the best and highest value?
00:31:00.640 --> 00:31:01.680
That's really all it is.
00:31:01.839 --> 00:31:11.039
There's no replacement, it's being able to put the highest value on that, which is a very common lean manufacturing print uh principle, you know, in the in this idea of waste and value.
00:31:11.200 --> 00:31:12.559
And so I can really appreciate that.
00:31:12.720 --> 00:31:28.480
With since you've worked with so many customers and so many industries, you know, which maybe industries or types of customers would you say are ahead of the curve on simulation, and which industries or types of customers do you think need it most that are not ahead of the curve with simulation?
00:31:28.880 --> 00:31:35.359
Yeah, I love this question because uh it kind of shows um yeah, yeah, it's the right question to ask.
00:31:35.519 --> 00:31:40.319
So the ones that are really ahead of the curves are the bigger companies with larger budgets.
00:31:40.400 --> 00:31:46.799
So everybody at the moment that builds larger rocket aerospace parts has to certify parts.
00:31:46.960 --> 00:31:49.839
Um, companies in the automotive industry.
00:31:49.920 --> 00:31:52.640
And I mean, I don't need to drop any names here.
00:31:52.720 --> 00:31:56.079
A simple Google search of people in the industry will know the big players.
00:31:56.319 --> 00:32:05.200
For them, it's easier to allocate budget to buy traditionally very expensive simulation software for metal additive manufacturing.
00:32:05.599 --> 00:32:27.200
And due to the nature of larger companies already or having in in a lot of cases dedicated simulation teams, they can invest in research and typically process cases that are, I want to say, two to three years ahead of the rest, you know, of average players that try to produce standard parts.
00:32:27.279 --> 00:32:42.559
Um, and that involves you know applying simulation top uh methodologies to investigate very specialized materials that are attractive to the requirements of uh aerospace application or defense applications or even medical applications.
00:32:42.720 --> 00:32:55.519
So that investment shows in terms of utilizing simulation and of course other research tools and experiments to understand how we can make the 3D printing process better, how can we make better parts?
00:32:55.680 --> 00:33:01.920
I want to say they're typically a couple years ahead of everybody else, and they're um utilizing simulation.
00:33:02.000 --> 00:33:05.519
And for them, I never need to explain to them why they need simulation.
00:33:05.680 --> 00:33:10.000
They come to me with questions and they know, so that they use it, they know why they want to use it.
00:33:10.079 --> 00:33:13.759
It's more a question of hey, do you think this is the right modeling approach?
00:33:13.920 --> 00:33:16.400
Or how would you model this specific problem?
00:33:16.559 --> 00:33:18.480
So it's a different conversation with them.
00:33:18.559 --> 00:33:19.920
Um, they know their stuff.
00:33:20.160 --> 00:33:27.920
It's more like providing consultancies for modeling approaches and ping-ponging ideas of how I would do certain things.
00:33:28.079 --> 00:33:38.240
Now, the larger group um that let's say um is a service bureaus that have outputs print still print complicated parts.
00:33:38.480 --> 00:33:48.799
Um, some of them have simulation tools, but for them, simulation tools and affording simulation tools, I think is often a budgetary problem.
00:33:48.960 --> 00:33:51.519
Simulation tools are not cheap.
00:33:52.000 --> 00:34:09.920
Um, again, we are trying to change that a little bit with having very low-cost products that makes it easier for shops that have one or two machines that um still have one, two, three failures per year that you can avoid by using simulation.
00:34:10.400 --> 00:34:23.760
Um, but I feel like there's this underserved market for you know um users that have machines but not the budget to buy fifteen to you know fifty thousand dollar simulation tools.
00:34:23.920 --> 00:34:37.920
So there they typically, you know, they can probably utilize a couple of thousand dollars, maybe several hundred dollars to buy simulation tools, but not you know, twenty-thirty thousand dollars per year to to to invest in that.
00:34:38.079 --> 00:34:51.440
So those users, smaller shops, um basically working on on producing parts, they typically understand why simulation is necessary and what it can do.
00:34:51.599 --> 00:35:06.960
But practically, and again, this uh my experience also having to do a lot of sales calls with with clients that uh are interested in simulation but not quite sure whether they should invest, is there's this debate on should I truly invest.
00:35:07.199 --> 00:35:20.320
So let's say you can provide a price that is attractive to them for simulation software, there's still a lot of hidden costs that they have to be mentally prepared to swallow, and that is you need somebody to train on the software.
00:35:20.559 --> 00:35:34.079
And no matter how simple you design your predictive tool, there there will be some basic training required and some general introduction in even with our co-engineer that can help with a lot of that.
00:35:34.320 --> 00:35:37.599
You also need to show people how you know to use that.
00:35:37.760 --> 00:35:41.519
So there's this kind of mental hurdle of how do we adapt this?
00:35:41.599 --> 00:35:44.079
Well, we need to train this person on it.
00:35:44.320 --> 00:35:51.199
We have that expense, even if the chop software is very affordable, we still um need to invest in that.
00:35:51.280 --> 00:35:57.119
And then we meet we need to make sure that the person preparing the build is actually using the software.
00:35:57.280 --> 00:36:12.559
And so all these factors, in my opinion, lead to the fact that uh smaller companies and medium-sized companies are still a little hesitant because of this adoption hurdle with the aforementioned reasons.
00:36:12.800 --> 00:36:31.039
Um now I still believe that in the future this is going to change when we look at how traditional simulation, structural simulation, um grew in in adoption in automotive or in powertrain industry.
00:36:31.199 --> 00:36:37.920
It started with the bigger companies and trickled down to smaller companies using simulation tools as well.
00:36:38.079 --> 00:36:43.519
And I truly believe this will happen in metal additive manufacturing as well.
00:36:43.679 --> 00:37:03.039
And our mission is to make that happen by having a very, very attractive pricing model that works even for small companies and making it simple enough to not having to worry about training that within one or two hours of introduction you can use this and you're good to go.
00:37:03.119 --> 00:37:17.199
And then having this AI tool that helps you remembering things when you need to and helps you double-checking your um your assumptions about the simulation to get rid of this scary thing of, oh wow, how do I use simulation?
00:37:17.280 --> 00:37:18.639
Do I have to be an expert?
00:37:18.719 --> 00:37:27.199
And I think that's essentially what we're aiming for, and the benefit for for smaller companies and mid-sized companies for adapting simulation.
00:37:27.360 --> 00:37:27.599
Yeah.
00:37:27.760 --> 00:37:33.440
And I I also think the more additive manufacturing moves into cereal production.
00:37:33.519 --> 00:37:36.079
So from I mean, that's already happening now.
00:37:36.159 --> 00:37:46.159
It it I truly think there is very good cases out there where larger companies move from individual parts to what I would consider serial production.
00:37:46.320 --> 00:37:52.639
And whenever you have that, quality uh control becomes a bigger discussion.
00:37:52.880 --> 00:38:11.920
And I see simulation as well as as part of this quality control pipeline that you make sure somebody at least looked into how well can this part, uh how well can this part be printed, and are there any issues that will reveal themselves once the part goes into service.
00:38:12.239 --> 00:38:19.199
I think that's so important when you think about not only, I mean, there's so much talk in defense and elsewhere about building up our industrial base.
00:38:19.360 --> 00:38:21.920
And that is not the five Fortune 10 companies.
00:38:22.000 --> 00:38:39.360
You know, that's the swath that you're talking about of small and mid-sized manufacturers where this idea of investment isn't just in the tool, it's in the personnel, and and you have that going in parallel with what you're talking about, where we're moving into serial production and being able to address both those needs.
00:38:39.760 --> 00:38:41.360
I really applaud what y'all are doing.
00:38:41.440 --> 00:38:43.679
You know, we've covered so much ground today.
00:38:44.000 --> 00:38:53.199
If you were to approach a manufacturing leader tomorrow and they were to take one thing away uh from this whole conversation, what would you want it to be?
00:38:53.440 --> 00:39:13.039
I would encourage everybody, whether it's an OEM, a service bureau utilizing manufacturing, or really anybody that is somehow in the vicinity or neighborhood of additive manufacturing, simulation can be a wonderful tool that can really help you to make your process better.
00:39:13.199 --> 00:39:19.280
It may seem scary, it may seem that you have to overcome using another software tool in your process.
00:39:19.519 --> 00:39:32.559
But if you truly dedicate some time to understanding what it is, what it does for you on the long run, it will save you time, it will save you, as a consequence, costs.
00:39:32.639 --> 00:39:42.000
And 100% I'm convinced that it will lead to improvement of your manufacturing process and the parts that you're manufacturing.
00:39:42.159 --> 00:39:43.920
I have absolutely no doubt about that.
00:39:44.239 --> 00:39:56.159
I think that is a flawless place to conclude this conversation, and it's a great takeaway for a lot of folks that listen to this that may be on the fence in terms of can this really provide ROI?
00:39:56.239 --> 00:40:00.079
And you've laid out such a wonderful business case in addition to the technical case.
00:40:00.239 --> 00:40:06.719
So I appreciate so much you spending some time with us today and and talking about simulation software and where it can add value.
00:40:06.880 --> 00:40:08.320
It's been a wonderful conversation.
00:40:08.559 --> 00:40:09.199
Thank you very much.
00:40:09.280 --> 00:40:11.199
Uh, also kudos to you and your team.
00:40:11.360 --> 00:40:13.280
I think you're doing a great job.
00:40:13.360 --> 00:40:18.079
Um, and I I like your dedication to podcasts because I like to listen to them.
00:40:18.159 --> 00:40:19.760
That's how I discovered you guys.
00:40:19.920 --> 00:40:24.880
And I think you had great guests that I've met at trade shows that I've spoken to.
00:40:24.960 --> 00:40:28.400
Uh, and I hope you keep this up because it's just wonderful.
00:40:28.559 --> 00:40:31.599
You know, when when I ride in my car, it's along a car ride.
00:40:31.679 --> 00:40:36.400
I put on your podcast, and hey, there's this guy that I met at the trade show that year.
00:40:36.559 --> 00:40:38.320
And uh let's listen to him.
00:40:38.400 --> 00:40:39.360
What does he have to say?
00:40:39.440 --> 00:40:42.320
And then it just, you know, for myself, it generates ideas.
00:40:42.480 --> 00:40:46.639
I understand how people think about this industry that I'm very passionate about.
00:40:46.880 --> 00:40:54.159
And thank you for sharing my perspective on what I would like to at the moment still consider to be quite niche.
00:40:54.239 --> 00:40:59.519
Um, it's not, you know, simulation is not everywhere, but I'd like to think it's growing.
00:40:59.760 --> 00:41:02.239
Users more and more understand what's going on.
00:41:02.400 --> 00:41:09.360
And I hope I can I could contribute with my insight to some of the myths around simulation and how it can be adopted.
00:41:09.760 --> 00:41:11.440
Well, you absolutely did that and more.
00:41:11.760 --> 00:41:11.920
All right.
00:41:12.159 --> 00:41:19.679
One of the things that obviously was key is Christian made a compelling uh point around the business case for simulation.
00:41:20.320 --> 00:41:29.039
Tell me about you know how you think that should be fitting into the conversations that our customers or our customers that are considering additive have.
00:41:30.480 --> 00:41:37.119
I think a lot of times we think about ROI as, well, I'm printing faster, so I'm saving time, and I can do more iterations.
00:41:37.360 --> 00:41:42.239
And that's true, but sometimes a bigger ROI is preventing expensive failures.
00:41:42.400 --> 00:41:50.960
And that was really his point about leveraging simulation from a business point of view, where every downstream failure becomes exponentially more costly.
00:41:51.119 --> 00:41:56.320
And this helps those manufacturers transition from call it reactive troubleshooting, why didn't this work?
00:41:56.400 --> 00:41:59.039
What should I change to proactive manufacturing?
00:41:59.199 --> 00:42:01.360
That is the key to scale production.
00:42:01.760 --> 00:42:03.280
You know, you make such a great point.
00:42:03.360 --> 00:42:13.920
And just to, you know, when we think about manufacturing best practices, the further a par a subassembly is down the line, the more you've got invested and the more expensive a failure is.
00:42:14.079 --> 00:42:23.039
And understanding where to apply simulation and other tools, um, essentially where are the failures most costly can be a tremendous guide.
00:42:23.280 --> 00:42:27.519
You know, he also, like a lot of our guests, spent a lot of time talking about AI.
00:42:28.000 --> 00:42:33.519
What stood out to you, though, in this conversation, maybe differently from the way AI has been brought up in our other episodes?
00:42:33.920 --> 00:42:37.199
I actually loved how he referred to AI as a co-engineer.
00:42:37.440 --> 00:42:45.199
And so when he was talking about its utility, we've had a lot of conversation around AI won't replace people, you know, it'll it'll help people be more effective.
00:42:45.440 --> 00:42:53.679
But he said this accelerating expertise, rather than pretending expertise isn't needed and outsourcing that to a tool is the big difference.
00:42:53.840 --> 00:43:05.119
And if you're a co-engineer, there implies some interaction, some insight, some exchange of ideas, and that's where he sees it being best applied in his own company, but in the simulation writ large.
00:43:05.360 --> 00:43:15.119
For a long time, the bottleneck is you would take a lot of expertise to know how to set up the simulation, and then three days of a PhD to interpret the analysis.
00:43:15.360 --> 00:43:21.679
And he said, you can't replace that interpretation necessarily, but oh my gosh, could I be more effective in setting it up?
00:43:21.840 --> 00:43:25.039
And could I at least have some analysis that is done?
00:43:25.119 --> 00:43:29.599
So now I am drawing insights into so what, what do I go do with this?
00:43:29.760 --> 00:43:32.880
Not this huge litany of issues, huge accelerant.
00:43:32.960 --> 00:43:36.159
And so for forever AI, I will be referring to as my co-engineer.
00:43:36.800 --> 00:43:41.840
Well, you know, that's exciting to everyone except real engineers like me who don't the thought of a co-engineer.
00:43:41.920 --> 00:43:44.480
Now I gotta argue with somebody about what the right answer is.
00:43:45.039 --> 00:43:52.079
Um who do you think ultimately benefits the most from what Novek is building?
00:43:52.480 --> 00:43:54.079
Yes, it's a great question.
00:43:54.239 --> 00:43:59.679
I asked them something similar, but I just said, you know, who do you think is benefiting the most from from simulation pure.
00:44:00.400 --> 00:44:02.000
And who should be that isn't?
00:44:02.159 --> 00:44:09.760
And he said, like many tools, when you have a Fortune 500 or a large aerospace company, they have a lot more resources at their disposal.
00:44:09.920 --> 00:44:13.280
So they will naturally be first adopters of this type of technology.
00:44:13.519 --> 00:44:22.639
But when you think about as a percentage of my investment, who is going to reap the biggest reward or suffer the greatest consequence, it's a small and mid-sized manufacturer.
00:44:22.800 --> 00:44:26.960
Every bet they place is hugely critical to their success.
00:44:27.119 --> 00:44:30.719
And so in his view, this is who should be adopting it.
00:44:30.880 --> 00:44:46.079
And if the cost is prohibitive, meaning I can't have two full-time PhDs on staff, tools like this can help lower that barrier to entry because you're doing some of that co-engineering work via a tool where before you'd have to have it via full-time FT FTEs.
00:44:46.400 --> 00:44:50.400
You touched on something there that I think is so important and often overlooked.
00:44:50.559 --> 00:44:55.199
And that is the smaller the company, the better your batting average has to be.
00:44:55.360 --> 00:44:55.599
Yes.
00:44:55.840 --> 00:44:56.079
Right?
00:44:56.159 --> 00:45:06.800
If I am, you know, I don't know, alphabet, Toyota, pick a pick a large company, how many bets are they making every month, every quarter, every year?
00:45:07.039 --> 00:45:08.880
Hundreds and hundreds and hundreds.
00:45:09.119 --> 00:45:19.920
If you are a company, you know, that is a you know, ten million dollar, hundred million dollar, five hundred million dollar company, you're making a relatively small number of bets.
00:45:20.079 --> 00:45:25.199
And the further you get down that size scale, the more you're counting on those bets to work.
00:45:25.760 --> 00:45:44.239
And I think his point around, again, what's the importance of investing where there are a lot of small and medium-sized business owners who I think would do well to consider, you know, your strengths are agility, ability to tailor a customer experience more easily for different customer types.
00:45:44.559 --> 00:45:52.239
But how do you how do you play to those strengths and then minimize the fact that my batting average has to be better than a lot of my larger competitors?
00:45:52.400 --> 00:45:54.880
And I think it's a great tool to improve that.
00:45:57.119 --> 00:46:06.960
We spend a lot of time talking about how to go from pilots kind of prototyping with additive into full-blown production at scale.
00:46:07.280 --> 00:46:11.599
How do you think simulation fits into that equation based on what he shared?
00:46:12.000 --> 00:46:16.639
Yeah, you know, when you think about what is required, production requires predictability.
00:46:16.800 --> 00:46:20.000
And predictability requires understanding your process.
00:46:20.239 --> 00:46:26.800
And a lot of times when you think about simulation, sure, it's failure prevention, but it's really a quality control tool.
00:46:26.960 --> 00:46:29.679
It's to help ensure that you have that consistent process.
00:46:29.760 --> 00:46:33.280
Um, and it's no different than inspection or process validation.
00:46:33.360 --> 00:46:39.039
And so I think that I have confidence that I understand my process and it'll work before I even hit print is huge.
00:46:39.199 --> 00:46:50.239
And we're talking about simulation, and it's used most often in the metal world, but that same mindset we have our VP here at B9 Creations always talks about building quality into each stage.
00:46:50.400 --> 00:46:52.480
That doesn't mean once you start building.
00:46:52.639 --> 00:46:54.719
That's even in the pre-build stage.
00:46:54.800 --> 00:47:07.360
And so I really appreciated his point around predictability and process and how this is such a vital part when you're having something that is very expensive, time and money, if you don't take the time to figure out what your failure points are.
00:47:08.159 --> 00:47:08.719
Agreed.
00:47:09.039 --> 00:47:10.480
Couldn't have said it better.
00:47:10.719 --> 00:47:19.280
Um, you know, this question we've started to ask if one of our listeners only took away one thing from this episode, what do you think that should be?
00:47:19.599 --> 00:47:23.679
Yeah, it was great talking to someone who uh doesn't live in the hardware and material space.
00:47:23.840 --> 00:47:34.079
Not those aren't critical, but we've talked a lot about uh with guests about the future of additive, you know, you need better technology, better materials, but really when you boil down to it, you need a smarter workflow.
00:47:34.239 --> 00:47:35.920
And that's what I would take away from this.
00:47:36.079 --> 00:47:38.400
And his work sits at the intersection of that.
00:47:38.559 --> 00:47:51.440
And I think it's if I have a smarter workflow that reduces scrap, makes me more money, helps improve repeatability, where I love that AI combination is then how can I democratize it and make it more accessible?
00:47:51.679 --> 00:48:01.760
So smarter, more accessible workflows, applying that methodology across everywhere you are leveraging advanced manufacturing tech is probably my one takeaway from that conversation.
00:48:02.159 --> 00:48:13.119
And I think that relates back quite a bit to the small and medium-sized businesses thinking about where they should be leveraging this to address some of the natural vulnerabilities of a smaller business.
00:48:13.280 --> 00:48:19.840
Um I wanna I want to touch on something that's coming up for everybody who is you know following along at home.
00:48:20.159 --> 00:48:23.280
So many of these conversations touch on the topic of workforce.
00:48:23.840 --> 00:48:31.360
And the lady behind the curtain, Nicole, had a great idea that we ought to do a series on workforce, like we did a series on Made in America.
00:48:31.440 --> 00:48:50.639
And so we're putting together, at this point, not quite sure if it's gonna be three, four, or five parts, but talking to policymakers, educational experts, industry experts around this topic because we hear it all the time, but so many times you hear you hear a word too many times, it kind of loses its meaning.
00:48:50.800 --> 00:48:56.239
We're gonna peel that onion and and dig deep into that in a in some future episodes and stay tuned.
00:48:56.639 --> 00:48:56.960
Love it.
00:48:57.119 --> 00:49:03.360
One of the biggest takeaways from this conversation is that simulation isn't really about software, it's about confidence.
00:49:03.599 --> 00:49:13.039
And confidence that your build will succeed before you spend thousands of dollars on materials and machine time, confidence your process is repeatable and as you move towards production.
00:49:13.119 --> 00:49:23.119
And ultimately, I think there's this idea that confidence needs to happen for additive manufacturing to be a reliable process rather than a series of expensive experiments.
00:49:23.280 --> 00:49:31.760
So we thank Christian so much for joining us and sharing his technical expertise and his vision for making this simulation more accessible to manufacturers everywhere.
00:49:32.159 --> 00:49:35.760
Thanks again for joining us on this episode of the Additive Advantage Podcast.
00:49:35.920 --> 00:49:45.760
If you enjoyed this conversation, be sure to follow this show on Apple Podcasts or Spotify, subscribe on YouTube, and connect with us on LinkedIn so that you never miss a new episode.
00:49:46.000 --> 00:49:51.440
Until next time, keep building, keep learning, keep finding your additive advantage.
