Inmox: Changing the Face of Industrial Monitoring & Maintenance

Zachariah Peterson
|  Created: September 5, 2023  |  Updated: July 1, 2024
Inmox: Changing the Face of Industrial Monitoring & Maintenance

Today we have a chat with Daniel Kagerbauer, CTO and co-founder at Inmox. This is a fascinating conversation regarding Inmox's aim to change how industrial maintenance is done and, while their initial focus is on gearboxes, the implications for their software and sensor developments may be broadly applicable across the entire industrial landscape. The two talk about monitoring industrial systems, materials challenges, industry-specific applications, and much more.

If you're interested in advancements in the industrial space, this is the episode for you.

Listen to the Episode:

Download this episode.

Watch the Episode:

Show Highlights:

  • Introduction to Daniel Kagerbauer CTO and co-founder of Inmox. Inmox is part of the Altium startup program called Launchpad
  • Inmox is currently developing industry ready prototype and moving towards heavy industrial applications
  • Commercial automotive and even the racing industry is a better fit for Inmox’s gearbox monitoring system
  • Daniel describes in detail what their product look like, from a sensor oil screw that can monitor the vehicle’s lubrication system, installing T-tube and wiring local ethernets
  • Data are being collected where the wear particles are present, and the oil screw with the lubricant have quite good access to essential stuff that needs measuring
  • Smaller systems are more automative focus while bigger systems are applicable for wind energy versions such as a helicopter, moreover safety is utmost important
  • Certification challenges can involve finances, redesigns and weight optimization
  • How does the real-time particle analysis works?
  • Daniel talks more about distinguishing between ferromagnetic, non-ferromagnetic, and the good old electro magnetism
  • Inmox is currently in negotiation with potential customers and doing interviews with mentors from different industries
  • Body vibration monitoring is more precise and reliable
  • Inmox longterm vision is to promote extended lifespan to machines and pushing mechanical engineering in a more sustainable path

Links and Resources:

Transcript:

Zach Peterson:

... the real-time particle analysis portion. So how does that actually work and does your sensor do it differently? Maybe can you talk about some of that without exposing any of your critical intellectual property?

Daniel Kagerbauer:

Sure. So yeah, there are these particle counters around which you put into also in the lubrication circuit and they basically count particles. They classify them a bit in size areas and they can distinguish between ferromagnetic and non-ferromagnetic particles.

Zach Peterson:

Hello everyone and welcome to the Altium OnTrack podcast. I'm your host, Zach Peterson. Today we're talking with Daniel Kagerbauer, CTO and co-founder at Inmox. Inmox is part of the Altium startup program called Launchpad and they have some very interesting technology that we're here to talk to them about today and this is all going to be very exciting I'm sure. Daniel, thank you so much for joining us.

Daniel Kagerbauer:

Yeah, nice to be here. Thanks.

Zach Peterson:

So if you could introduce yourself and your company, Inmox. Particularly, what does your company do and what's your technology solution?

Daniel Kagerbauer:

Yeah, okay. Again, nice to be here. I'm Daniel, CTO and co-founder of Inmox. I founded the company together with a colleague. We both studied at Technical University here in Vienna. I have a physics PhD and the colleague has background in mechanical engineering. So we put together two different fields, but especially for this application, that's quite an important thing as we learned. So we jumped off and our goal is to change the way maintenance is done. So, we saw in different applications that maintenance is done more on a calendar based way. So we said, well that can be done better. And we thought new technology can help with that and we set us the goal to not only do just software based system, but we developed a sensor ourselves because we also saw the need of a more direct parameter. So when I say maintenance, it's at the moment focused on gear boxes. Gear boxes in general, but more specific on big industrial gear boxes, like there are in the wind turbine or in a helicopter or in big production lines for steelworks, automotive production lines and so on, so things which keep running.

Daniel Kagerbauer:

And to monitor them, we developed the sensor system, which we implement in the lubrication circuits. We need access to the lubricant and in this lubricant we detect different wear particles and characterize them, which gives us the opportunity to pinpoint down where such a defect or such a damage starts so that we can act. So it's like the information for a customer is then, "Well, the bearings don't look good, maybe have a closer look at them. Or they start to degradate you need to change them. Or it's coming from a tooth, so there needs to be for next repair, for example, to do something with the tooth wheels." So just to get more information about your system, about the risk potential so that you can act accordingly and not just go there, have a look, go back, take the wrong equipment, wrong people.

Daniel Kagerbauer:

So really to be informed and maintain your equipment in a good and healthy way. Also in a sustainable way, especially for wind energy. If you think wind energy is a sustainable energy production, I think don't have to argue about that, but sometimes it's not used sustainably. If you go there and change the gearbox very often, there are 700 liters of oil in it. You have to change that as well. So, there's going to be a lot of things going on around it. So, using it sustainably is also something which we're quite keen on and then hopefully we can change that.

Zach Peterson:

Yeah, you mentioned a few different markets there. Large industrial systems, I think that makes sense, like big production assets. And then wind turbines, obviously these massive structures that of course are going to have this big gearbox as well as many other mechanical parts and electrical parts I'm sure. I would think automotive would really be the natural place where you guys would fit in. What does your penetration look like into the automotive market?

Daniel Kagerbauer:

The other thing is our company's now just two years old. We are still in the phase of developing an industry ready prototype. So this is our next big goal. So, we got there some parts now and we are planning to implement that for test applications. But in this stage, the equipment is just quite expensive and therefore it doesn't make so much sense to jump into automotive, which is a very price sensitive market. So we decided to go with the heavy industrial applications. But also as a first look into automotive, we defined heavy duty applications like construction vehicles, trucks, where there again they should run. If they're running in remote areas, then even better. If they're controlled remotely, then even better. You need some sensorics to monitor it and there get feeling and get information about these markets, get information about special requirements you have there.

Daniel Kagerbauer:

Automotive there, you need certifications and so on to learn that. And during that also maybe we learn that we don't need the full scope of the sensor system. Maybe we can scale it down a bit for automotive because we measure the important things anyway and with a smaller sensor then we are getting into an area where price is not such an issue and we can tackle automotive. So this is kind of the scope, but this is more a long-term goal for us.

Zach Peterson:

So it sounds something more like commercial vehicles maybe a better target before you get into the general consumer automotive market.

Daniel Kagerbauer:

Yeah, exactly, exactly.

Zach Peterson:

Yeah, at the freight level.

Daniel Kagerbauer:

Yeah, especially then if you think about autonomous driving with these vehicles, sensorics is quite an important thing to have it. And if you know something about your vehicle in a health way, like what exactly is the status of your vehicle? Another approach can be via something like racing. The test in racing quite the new high tech stuff. They're not very price sensitive if you pitch them that you've given an important contribution to it. If you think about Formula One, if you can tell a team that they know exactly if they can use their gearbox one race longer, that makes a big impact and can be a pretty good test field. They're used to tests, they are top notch, and then used to new technology.

Zach Peterson:

Well, I'm sure for the racing guys they could then, if they wanted to, correlate that to performance. They could really say, "Hey, yeah, we see that maintenance is going on but it's really not impacting performance so maybe we can get even a couple more races out of it."

Daniel Kagerbauer:

Yeah, exactly. You learn a lot out of that. This is for us as well quite the goal to get them. We know our sensor system won't be the only one in any drivetrain to be fair, but we see it as an important add and that the most powerful thing comes from combining all these data. As you said, if you know their performance data, if you know your turning speeds and loads for the gearbox and then you can correlate that with what's happening in the lubrication system from wear perspective, wear particles are coming from, that can be a very big playground for that. Data analytics, very specific data analytics as you said, if you then know when I take that turn so fast then the gearbox gets damaged for example. I'm not sure if that's possible, but still you can imagine in this direction.

Daniel Kagerbauer:

Or for a gearbox, for a gearbox in a wind turbine, when there are special wind conditions you know that damages starts, then you can adjust the wind turbine. You can change the pitch. You can change the yaw. And if that helps you to keep your equipment longer alive, then that helps you for your revenue in the end as an operator.

Zach Peterson:

So what does this product look like? Physically what does it look like? Does it live inside of a gearbox or does it connect in through tubes? And then how are you getting the data out? Is it wireless? Is it wired? Does it send it over the internet to a cloud platform?

Daniel Kagerbauer:

So, we try to be there as agile as possible to adapt to use cases. The general thing, what we need is we need to be with our sensor to be in contact with the lubricant. So, we need access to the lubrication system and there are basically two entry points which we are planning at the moment. One is the oil screw, so just to change the oil screw to a sensor. So, our sensor will look like a cylinder with a certain diameter and then a certain length. So, at the moment we're at an M40 screw and have 150 millimeter length, so this is plus minus at the moment, which is a bit big for oil screws, but gearboxes in wind turbines are quite big. So maybe there we are happy with the oil screw.

Daniel Kagerbauer:

And the other thing is to just change part of the tubing, so to put some additional tube in front of the filter. So, put there a T-tube and attach to the one side of the T our sensor. The idea is to change the lubrication circuit as little as possible so that we don't change flow velocities and so on so that we don't impact the system with our sensor, but just be the silent listener or measure system to it. So this is the mechanical application, again, depending on the application. Just size wise, is a gearbox in a steelworks something else than a gearbox in the wind turbine? So, we try to be there as adaptive as possible to the application.

Daniel Kagerbauer:

Then on the other side, how to implement in the software system again depends very strong on the application for wind turbines. For example, we're evaluating at the moment to send data with the SCADA systems, so with systems that are in place already, but there are some downsides with it. At the moment, we're also looking more on onshore wind turbines. So, they're just using a GSM data module to send it into a cloud. Can be a good way to go for first tests for example. So, we set up already a data stream from our measurement PCP to our cloud. So just transform data there and get their first way of showing the data, showing a trend. So getting the whole time series of it to really measure what happened in the last, I don't know, weeks and can see trends changes and then analyze that.

Daniel Kagerbauer:

So more in a viewing point, but we are planning to build that up to have APIs that customers can include that in their existing systems because there are existing monitoring systems in place so that we can implement that there. And there's the other quite completely, no, we never want connects to the internet part where they say, no, no, no, everything has to stay local. There are some wiring around there. So, we go wire local ethernets and send it to a data storage in the production line, for example. But there we can be flexible. Our team is now with us two co-founders and we have four more people on board with two tech guys, a electrical engineer and the software engineers. So, we're there quite broad and we can adapt to these things, but well, we need to know and there are some test fields in the pipe where we know these things already and then work on implementing it.

Zach Peterson:

When you're describing the sensor, I would've assumed that it literally lives inside the gearbox. So, you would have to design it so that it's taking up space in the gearbox where it's obviously not in the way of any gears or any other things that are in motion because obviously that would give you access to the lubricant. Why didn't you go that route? And what are some of the mechanical challenges with developing a product like this? Because if it's a screw, it's obviously not a planer device. Whereas if it were to go into the gearbox, it could be a planer device.

Daniel Kagerbauer:

Yep. The good thing is with how a gearbox is designed, it's designed that the lubricant really not only lubricates but also cool, sometimes that depends on the application, but it's pumped around. So you get a good chance in seeing all the wear particles by monitoring this lubricant because it's pumped around, it should enter or it should be at all the spaces where there are high frictions and high forces. So where the wear actually starts. So, that can be there flushed away more or less. And for us it was the idea to stay in front of the filter because a filter is designed to get out all these wear particles. So we want to be in front of that, that we can measure it.

Daniel Kagerbauer:

But there's also another way to just be in the oil sump. So just on the bottom of such a gearbox. And with the oil screw, that's most of the time at the lowest point. So, you get a high chance to see all the wear particles because they're moving down there. To actually add something to the gearbox or in the gearbox itself is quite complicated because it's hard enough to get people to that you can add something to their lubrication system, but to change actually the setting of the gearbox because the gearbox is a major and very important part of the drivetrain, so people want to keep it intact and don't want to change stuff around it. And with the lubricant, we have quite a good access to the things we want to measure.

Zach Peterson:

Okay. So once you put this onto, let's say a gearbox in a wind turbine, I'm sure the vibrational environment in a wind turbine is going to be very different from let's say Formula One racing or an industrial system or maybe commercial freight. How does that impact your results and do you compensate for it or does it affect the design of the product in any way?

Daniel Kagerbauer:

So the good thing is that the measurement principle itself is not affected by it. So, we are measuring magnetically and we're detecting these materials and these materials. Maybe Formula One is different because they use different materials. But in general, if you think about mechanical engineering, there are similar materials around. So by us detecting it, that doesn't make any difference if it is in a truck or in a gearbox of a wind turbine. The whole environment, you're totally right about that. So this is something we will investigate in the experiments coming now in the next month when we start with our pilots because we can't mimic that really good. So we need to get out, see what's happening there. As I said, the measurement system isn't, so the measurement technique itself isn't influenced. So this is a good thing. The environment, we built the prototype quite big and massive, so we shouldn't have too much problems with the electronics there just to be on the safe side and learn from the test applications.

Daniel Kagerbauer:

But this learning then will be put into a direction of a serious ready product where probably there will be different versions of it. The wind energy version quite bigger because the system is bigger. And then a smaller, more automotive focused version with the specific... And also if you think about the whole certification processes, that will be different in the different industries. And if you think now about aviation, about a helicopter gearbox, the biggest challenge there is to go through the certification process. So, it is an interesting market for us because safety comes there on top on your profit of running your system because you want to monitor your helicopter because the helicopter, there shouldn't be any gearbox damage. That's quite important. But of course the whole certification is huge to get into it.

Zach Peterson:

Yeah, talk more about the certification aspect because I think this is something that not all designers have to deal with or maybe they don't deal with it until they change jobs. They go from one industry to another and then suddenly they realize, oh, I have to learn these new safety standards, I have to learn these new design practices to ensure we can pass certification, things like this.

Daniel Kagerbauer:

Yeah, the easiest thing I think where everybody's in contact with it, these ICs with the specific automotive certification on it. That's the things you use or the military certification, that's just then parts you use. To be honest at the moment, you're also trying to be fast and not taking that too much into consideration because we need our prototype. We need to work on a prototype and it needs tested. And for the testing it's fine that it's not certified in the hallway. Wind energy is not so strong on the certification side, so there I think it's quite fast to get into it. It's combined with a lot of money that you give it to somebody who looks on it and then says, "Well, that's fine for this area." But if you look into aviation there a lot about this failure critical stuff, maybe having components twice that you make sure that nothing stops in there. So this is going to be a total, to be honest, a total redesign of the whole thing where we have to focus on to get it into the different industries and that needs to be investigated thoroughly.

Zach Peterson:

Aviation is an interesting one because you've got a tough mechanical environment in terms of thermal cycling and then vibration, but then you also have some particular electrical requirements, particularly with ESD. So, I'm sure that it changes the design approach once you go from automotive to aerospace.

Daniel Kagerbauer:

Yeah. And then add on that the weight criteria. So there it's getting, again, that you need to... It's not like well just build it bigger that it can withstand more. That doesn't go with aviation.

Zach Peterson:

Sure, sure. So I'd like to maybe shift gears, no pun intended, into the real-time particle analysis portion. So, how does that actually work? And does your sensor do it differently? Maybe can you talk about some of that without exposing any of your critical intellectual property?

Daniel Kagerbauer:

Sure. There are these particle counters around which you put into also in the lubrication circuit and they basically count particles. They classify them a bit in size areas and they can distinguish between ferromagnetic and non-ferromagnetic particles. That gives you some information. But we learned also, by interviews with industry partners, that it's not enough information that the valuable information is missing. So what we are... And also, sorry I have to add that as well for the particle counters, you need moving particles. So, there it wouldn't work to just put the sensor in the oil sump. For us, we don't need the moving particles because our magnetic measurement technique stimulates the magnetic field by itself. So, you measure either in the oil sump or in the pump line, that doesn't make any difference for our system. And we take this characterization step further.

Daniel Kagerbauer:

So, we also detect particles. We can also count them. We have some size measurement in it, but the really new thing is this classification of the material. So we can tell by using a material database which we set up, this is this kind of steel, this is a hardened steel, this is not hardened steel. That tells you something about the wear from the tooth wheels if it's coming from the surface or are you further in. Or if it's brass or bronze, especially brass and bronze are hardly detected by particle counters. And these are materials in bearings which is quite important to measure them. So on the one hand, these very focused characterization of the particles, but we also see ourselves in helping the customers really to interpret it. We don't want to put their box and then you get there out some signals and then deal with it.

Daniel Kagerbauer:

But we see ourselves really in this position to understand the system, to link together the things we measure, and then give them just to interpret it to say, "Well, something with the bearing is starting and now you have problems with tooth wheels," for example, which you see from the data. And then this in some kind of review for maintenance crews. Then the maintenance crew going there or going there for decades, they know the systems as well, but if they get this additional information to look closer at specific positions in the gearbox, they just can look more efficiently and find the damages and do the changes.

Daniel Kagerbauer:

And we see ourselves in also giving this and start learning on it and really building damage models to understand because that's always a tricky part when you detect where in the lubrication system to tell something about when the damage is, when that damage really starts. To get to the point where we are able to do a really lifetime estimation to tell whole gearbox that's now around two years to go, but you have to look into your bearings that you just have one year left, something like that. This is future music, but we see ourselves going in this direction. And for that, as we talked earlier, it is important to see as well other parameters because they can tell you something about the whole system.

Zach Peterson:

You brought up determining which material was making up the particles. I would've thought that this was just size, shape, density and then throw some AI at it and then you're able to infer, oh okay, this particle flows from the bearing or this particle flows from the gear teeth or something like this. But I think you said you're actually detecting which materials or are at least distinguishing between ferromagnetic and non-ferromagnetic.

Daniel Kagerbauer:

Yeah. So it's more than ferromagnetic and non-ferromagnetic. We're using all good old electromagnetism with the magnetic properties. So the good thing is that the magnetic properties can be linked to the material quite good. So for us, you can't tell that for everything, but for the materials we're using on and with the material database, that can also be expanded. So if there are new materials used or interesting, we can add them to the database and distinguish them from others.

Zach Peterson:

So someone who wants to implement this in their system, they would at least need to tell your software, these are the materials that are present in the gearbox, whether it's bearings or teeth or something like that?

Daniel Kagerbauer:

Not necessarily that depends strongly on the application. So mechanical engineering is a quite conservative field. So if you look at the average or more than average application, there are norms in place. So, there are the same materials used in most of them. So for most applications, our sensor is a plug and play system because we know with a high percentage what's built into the gearbox. If this is a special application, for example, Formula One racing or whatever niche application where you use the new kind of steel because the properties are better, then we would need to know that. But these people tend to know that they use something not ordinary and then we can measure that, characterize it, and put it in just on the software side. So there wouldn't be any change hardware wise, but we can do that in an update to detect this material additionally.

Zach Peterson:

Okay, interesting. So, I guess now we can ask what kind of impact is the product having on customers? And what are they telling you about their results?

Daniel Kagerbauer:

So, we don't have any products yet out in the wild. So we are negotiating with pilot customers at the moment, but we did a lot of interviews with potential customers and mentors from different industries. And there most of them told us that it is important information, that they sometimes struggle by interpreting data, that they have inconclusive warnings, which they have then to do by their gut feeling, go there, not go there. So in general for different fields, there are experts, they know their systems very well. But well, if they retire then that knowledge is not there anymore. So you see in a lot of industries as well a push to digitalization so that people just see they want to monitor their equipment in real time. They just want to know what's going on there and they're quite happy talking with us and then doing that because it's a risk monitoring and the detection, the characterization of the wear particles.

Daniel Kagerbauer:

Basically everybody was quite interested in it because they said that can be the linking information or the missing information, giving them the opportunity to make more informed decisions. Everybody's now interested in trying it, so we need to get it out and try it and then next round we compare it with other data sets.

Zach Peterson:

Yeah, you talked about experts basically going on their gut feeling or their very extensive experience to determine when to implement maintenance. I even seem to recall reading a paper from several years ago that looked at acoustic assisted predictive maintenance. Basically they're listening to the sound of different parts of the system to determine when they're going to do maintenance. And if they see an appreciable change in the spectrum, so the type of noise and then the level, then they say, "Oh, okay, now it's time to do maintenance." Is that something that you're trying to replace or can you compliment that?

Daniel Kagerbauer:

Yeah, so the whole vibration monitoring in general, so there are two systems I would say. The one is just looking at if it's moving more than before, then it's bad. But this is quite a late detection system I would call it. But there's also just the very precise analysis where you do this body vibration monitoring where you basically do then an FFT and look at Inspectra and how they move. This is very accurate, yes. But there you are very reliable. You'll rely on knowing the surroundings. So that now depends how much vibration is coming from outside, where is it positioned, and then you need an expert to interpret it. And on top of that, these systems don't work for planetary gears and slow rotation gears. And in wind turbines, both are present.

Daniel Kagerbauer:

So planetary gears are not really good monitored by this vibration monitoring. You can do this late stage vibration monitoring, but well, you can also put a camera in front of the wind turbine and see if the rotator moves too fast. This is a late detection, but we see ourselves basically giving the information these systems give, but more accurately because we have an indication which is actually going on that a damage will start in the next time interval.

Zach Peterson:

Yeah, I see. Based on your closer to real time analysis, you're able to say, "Okay, now you should probably schedule this more intensive investigation that can only be done let's say once every six months or once a year." And then it really does require specialized equipment, more expertise to interpret the results, things like this.

Daniel Kagerbauer:

Yeah. And you can also plan, for the example of wind turbines, you can plan on tower repairs because if you do a full gearbox change, you need a big crane. And I talked with a colleague, and especially in the US, apparently cranes are not so easy to get. So, people are really looking to do on tower repairs. And the smaller the damage is, the easier you can do an on tower repair. So if you know that early on you can repair and don't need to wait for a crane and that can be months and then your system is just not running because it's broken. So to plan it, even in a not normal maintenance but do a smaller maintenance, you know where to look, you know what to change can make a big impact.

Zach Peterson:

Okay. That's all very interesting. So maybe just before we close, you can tell me what is the longer term vision for the company? And what is the longer term vision for the product or the suite of products that you want to develop?

Daniel Kagerbauer:

So we see ourselves, as I said it before, as this help for the customer to really giving this interpretation and the longtime vision is to have this lifetime extension. So to really build up damage models, know the systems, can tell specifically for parts or assemblies, how long their lifetime will be or expected lifetime will be, and then you can act on that and change on that. To achieve that, while in our heads there are quite some interesting parameters we want to monitor and maybe another sensor system will come handy for that and we will develop in this direction. So, we will look into what additional things would help, not only rely on existing data but also think this often hard step because it includes hardware to really develop something new if it's necessary to help us to get this lifetime extension and really push mechanical engineering in a more sustainable way, especially the maintenance because that's not done sustainably in all areas. And then we see ourself with the lifetime extension to help there and to plan the maintenance and do it sustainably.

Zach Peterson:

Well, I think that's a great vision for the company. And as it unfolds, we hope to have you back on in the future to discuss all these developments.

Daniel Kagerbauer:

Yeah, it will be great. Thank you.

Zach Peterson:

Thank you so much for being here. To everyone that's out there listening or watching on YouTube, we've been talking with Daniel Kagerbauer, CTO and co-founder at Inmox. If you are watching on YouTube, make sure to hit the subscribe button. You'll be able to keep up with all of our tutorials and podcast episodes as they come out. And last but not least, make sure to check out the show notes and of course, don't stop learning, stay on track, and we'll see you next time. Thanks everybody.

About Author

About Author

Zachariah Peterson has an extensive technical background in academia and industry. He currently provides research, design, and marketing services to companies in the electronics industry. Prior to working in the PCB industry, he taught at Portland State University and conducted research on random laser theory, materials, and stability. His background in scientific research spans topics in nanoparticle lasers, electronic and optoelectronic semiconductor devices, environmental sensors, and stochastics. His work has been published in over a dozen peer-reviewed journals and conference proceedings, and he has written 2500+ technical articles on PCB design for a number of companies. He is a member of IEEE Photonics Society, IEEE Electronics Packaging Society, American Physical Society, and the Printed Circuit Engineering Association (PCEA). He previously served as a voting member on the INCITS Quantum Computing Technical Advisory Committee working on technical standards for quantum electronics, and he currently serves on the IEEE P3186 Working Group focused on Port Interface Representing Photonic Signals Using SPICE-class Circuit Simulators.

Related Resources

Related Technical Documentation

Back to Home
Thank you, you are now subscribed to updates.