top of page
  • Writer's pictureThe Solar Journey

Episode #017 - Jürgen Sutterlüti

Updated: Dec 16, 2022

Episode #017 with Torsten Brammer & Jügen Sutterlüti:

In episode #017 Torsten Brammer and Jürgen Sutterlüti talk about the optimization of solar power plants by using Edge computing and cloud-based processing.

Jürgen Sutterlüti is responsible for the energy business segment and cloud platform development at Gantner Instruments, a global leader in modular and flexible data acquisition systems. He gives an overview of the key challenges in operating solar power plants and describes how they can collect and utilize their data in the most effective way. Learn more about what Edge computing really means and what else is required to take solar to the next level 🚀.


Mr. Juergen Sutterlueti (Jürgen Sutterlüti) leads the Energy Business Segment and Cloud platform development within the Gantner Instruments Group with a strong focus on monitoring and controlling energy assets using Edge computing devices and cloud-based post-processing by driving the development of application-specific services for PV, Asset monitoring, test setups, and structural health monitoring with a clear focus on scalability and the lowest running cost, as well as new data logging and control systems for energy applications, e.g., batteries.

Data-driven decision-making with mechanistic and machine learning-based modeling is the latest competence he established at Gantner Group, with a wide implementation at different company segments in parallel to establishing inbound marketing across all sales teams globally.

Before joining Gantner Instruments in 2014, at Oerlikon Solar and Tokyo Electron Solar, he was responsible for optimizing PV power plants to enable the most efficient transformation from "Wp to kWh," and he led the PV Systems Group, which executed global Outdoor Performance assessments, BoS cost reductions, and LCoE analysis, with a focus on Balance of Systems (BoS) modeling and identifying BoS cost reduction potentials, including market implementation, as well as the performance optimization and characterization of PV modules in different climates and the related long-term behavior and technology benchmarking.

These findings were implemented in optimizing simulation and performance prediction tools for PV Modules and PV power plants, as well as LCoE sensitivity assessments.

Juergen Sutterlueti studied Power Electronics and Systems Engineering in Austria and Switzerland and holds an MBA in Entrepreneurship from the University of Liechtenstein, HSG St. Gallen.

Show Notes:
  • (10:32-12:05) What does it take to be successful in the #PV market? It’s always a balance between research and operational business.

  • (25:00-28.33) Edge Computing is a buzzword, but what does it really mean?

  • (52.44-55.07) #Drones for IR or EL imaging are a big topic, which will grow even more in the future.


[00:00:17.490] - Torsten

So hello everybody, and welcome to a new episode of The Solar Journey. My guest today is Jürgen Sutterlüti. Welcome to the show, Jürgen.

[00:00:29.460] - Jürgen

Hello Torsten, great to be here.

[00:00:31.800] - Torsten

Yeah, thanks for joining. So, let me briefly introduce jurgen works for an Austrian company called the Gantner, and he's the vice president in the energy segment and he's responsible in that role for cloud services and marketing. So his business units provide services for monitoring and controlling energy assets. So Jürgen helps to optimize the operation of solar power plants and battery parks. And the methods used for these services are edge computing devices and cloud based processing. Before joining Gantner Instruments in 2014, he worked for Early Consola and TokyoElectron Solar, where he was responsible for optimizing PV power plants. So Jürgen has been in the solar industry for quite some time, and power plants were always on his mind. His background, Jürgen studied power electronics and systems engineering in Austria and Switzerland, and he holds an MBA in entrepreneurship from the University of Lichtenstein and University of Sank Island in Switzerland. So, Jürgen, when and why did you get started in solar?

[00:01:57.490] - Jürgen

Well, so great, welcome from my side also to everybody and all the listeners. Well, when I did start in solar, it was I think roughly 2003 when I did my masters on the university where it started with renewable energy and all the IPCC reports where they are starting to get a bigger topic. And there were also a few niche applications for PV and a few companies around who thought there is some potential in solar. And then we did learn that. I did also do a lot of courses for renewable energy systems and also see the global energy change. The BP reports were a big thing at that time. You have to understand where all the oil is coming and I try to understand how this system works. And it's a hot topic upfit today. Maybe it's switching out to gas, but still it's oil and coal and gas dominated system. But it was clear that the system has to change because everybody knows that PECO is over and already passed. So people are looking to keep their power, but still move to some new applications. And then I had the chance to work in the universities to set up a renewable energy institute where we did a lot of PV.

[00:03:20.290] - Jürgen

And at that time in the rental valley, there were also a lot of activities for starting coding activities for solar industry, which was early in solar. Obviously coming from the Big Oerlikon Group has a perfect coding experience. A lot of people know it from Balters and so on. And PCVD was needed and then we could start and work a lot out for new PV, feasibilities and so on. Also inverter manufacturers, transformerless manufacturers where they're coming to the market. So we detest and analyze and model a lot. And that was quite intensive and also fun to see the different journeys, where the different players are going and that build a lot of competence. And it was a very good self financed institute. So we had more than 50% industry financing. So the demand was there and this brought a lot of foundations there. And then you start to do papers, then you start to meet people in the industry. That was one of the first phases entering solar.

[00:04:36.640] - Torsten

Excellent, excellent. So then from the university you switched to Oerlikon. What was your initial job at Oerlikon?

[00:04:45.940] - Jürgen

Yeah, I mean, the timing was perfect because there was a demand for doing something when a TV module was produced out of that EndToEnd turkey production line early in the beginning has not a concept that they do turkey. They wanted to sell just equipment. That's the same trend as the people are doing now. Also again and so on, the PCVDs and Chai machines were more or less they're a key thing. And then they decided to build up a turkey thing. And then you need somebody who understands what should the module look like, what is coming out in the outer conditions and how do you link outer measurements with flesh measurements? And building up their team and working also very close with the researchers in Nashatel. So we always had access to very bright minds where it could work with them, test their new sales modules, testing the latest stuff pilot line. I think you know that feeling and you know that journey as well. And so we try to understand what's coming out in the real world and there are a lot of discussions why it's not as it is in the flesh. And that was always a motivation for me to understand that, even against resistance, because standard test conditions are for indoor and outdoor, it's something else.

[00:06:18.250] - Jürgen

And on top of everything, it was very clear that it's kilowatt hours per square meter. And this is what makes the win. And this win we tried to really write the Wave with Sin film because Crystalline was very expensive and so on. It's expensive again today, so that's also a driver for new technologies. But you also had to mount that module. So we had working a lot of system designs. What is the PV module looking like? So a lot of trends for solar use, for example, we also initiated and did their things, a lot of power plants. And this was more or less the early journey. This was then getting bigger and bigger. So we had to convert customers in the home market. So we test power plants across the globe, from Alice Springs to the US power plants next to each other, so we have local data for India and so on. It was also fun to control and set up everything. And there is this first step of collecting all the data because you had to have optimized reporting, otherwise you spend too much time on Excel files and so on. And you also have to select the right database, which was then a good starting journey and learning effect for my work at Gantner.

[00:07:54.710] - Jürgen

But coming back then to the power plants, it's always then that you want to compare. You always have to compete as a new technology versus established ones. And if you don't show the benefit upfront, you are not there. So it was very intense and also very strategic that you have to show track record, and you cannot accelerate track record in the field. So you have to install early the stuff. And we did that at several locations worldwide.

[00:08:27.490] - Torsten

Excellent. So lots of engineering that drove your initial career, but also with the focus on the economics, the profit from solar power plants, and you did an MBA on top of your engineering degree, or actually two, if I read correctly.

[00:08:46.910] - Jürgen

It's a combined one.

[00:08:47.980] - Torsten

It's not just one, it's just one combined, yeah. What was the driving force? Was it worth it, and would you do it again? Or should you have gone business first? Should you have skipped engineering when you look back now?

[00:09:02.590] - Jürgen

No, I think the people in the market are very helpful when they can understand both worlds. That's also my interest. For me, it was also very clear that I'm not a full 100% researcher, but I want to bring something to the market and identify new trends and use that technology to create value. This is what drives me. And when you see new ideas, when you see a new energy generation source, I wanted to test that. That's why I also went to wind industry and developing converters and understand what happens when a DC AC converter blows up and all that stuff. So I was really onsite repairing that stuff, the cooling liquids and accidentally hitting the shutdowns and so on. So that's all experience which you have to have, because when you do some strategic things and more business related things, you cannot run the world by Excel or business plans. Somebody has to do it at the end. And when you can judge that what it really means, I think you can work with the team in a different way. And I think that's the way how I want to work with people and also how we have to have project development nowadays.

[00:10:31.590] - Torsten

Excellent. So what was the key thing you learned during your MBA which you didn't learn during your engineering course?

[00:10:39.860] - Jürgen

Well, I think it's very important that you can communicate your idea, what you want, and that you have to spend a lot of time to simplify it down for the right audience, and that you have to be quick adapting to the audience, because that changes over time and over the day. And if you don't understand what you are doing, people will discover that very quickly. So don't talk about things or be open when you don't know things and be honest. So that's a little bit. My main takeaway the NBA was about entrepreneurship because I want to understand and learn that as a technician you saw that a lot of startup was a hot topic and I wanted to understand what are the rules and how you work with that VCs and all that stuff. And that was a very good background that you have an understanding about how it works. Also, digital marketing is something what we touched here. So it was a lot of topics which I understand at the time very well. And it was one of the first entrepreneurship studies in Europe with excellent profits. So we benefited there a lot and that's helping me up to today that you have a little bit of understanding how the business world works.

[00:12:05.500] - Torsten

Excellent, thanks for that. I'm sure that many listeners consider moving on from one direction, engineering or business to the other side. And this could be helpful input for some of the business. Hey, a few words on Ghanner. So how many people work at Gantner? What are the other segments? Roughly revenue, global footprint? Could you share some information on Gantner?

[00:12:37.690] - Jürgen

Yes. So Gantner Instruments is an Austrian based company headquartered in Fallback, so close of Lake of Constance and in the Montavon. Nice work. Exactly. Blue sky today is no on the mountains, so perfect work place, but it's a very distributed and flat company and it was found in 82 with a lot of different applications from access systems, from RFID detection system and data acquisition. And the investors at that time, they had said too many different segments does not make sense, they have to focus. And then the data acquisition system was one which was one up for sale. And then our current CEO, Mr. Bernardanal and the colleague, they bought that out and then he was running Instruments, which is then doing data acquisition and control. And one application in that area was mobility, automotive combustion engine nowadays EV testing, battery testing, the other one is aero and space, so rocket launchpad pads and so on. So also very dynamic growth now in the Space Coast. So we have clients in that Boeing Airbus area which brings all of the mindset of a company into the product because even when you're in a small valvey, you have to delivering for the world.

[00:14:15.750] - Jürgen

The second thing is that we have then also condition monitoring and civil engineering applications for railway and so on. German railway, for example, are big customers from us. And then we have energy. In between energy, we're doing wind, solar battery and also hydro power plant monitoring. And PV is the latest add on on the gross journey because PV started quite small and is now 30% of the turnover already. Overall we make 100, we make 25 million plus and we have more than 100 people across the world. We have offices, own offices from San Diego to Singapore, to in China, Paris, Stockholm, Germany. Two in Sunitz and in Nunberg and headquarters in Austria. So it's like a very flat organization. RNT is done in the headquarters and sales and business processes and talking to the people is very decentralized. That helps us that we get very early the trends from Asia, what they at the end will be implemented or demanded in the US. And Europe is in the middle. We have more or less twenty four, seven of support with all the teams, which makes us accessible. And the focus is that you and that was not just because it's now hip.

[00:15:59.140] - Jürgen

It's at the beginning that you listen to the customers and deliver rapid prototypes and feasibility with them. So we go to journey with them. It's partnership, not just a vendor. And that brought us to a lot of cool projects and that's a lot more or less the spirit how you grow in the new markets. For example, we are also doing in the energy segment fusion energy, which is always delivered in the next 30 years something, but it's shifted all the time. But you know that fusion rector in Greyswell, North Germany, all the data acquisition comes from us and even the note of us, the emergency stop is controlled. This gun, this is like a flagship project and you don't get in there if the quality is not good, period. And due to that, we are also working in Ether and all that stuff. And the same examples are applicable for a lot of different applications. So we have all the automotive players which are doing their testing with Gantner really globally. So you have to have people in Michigan, in Detroit, you have to have people in Munich, in Stukart, which they are talking to the customer, working with the customer.

[00:17:22.460] - Jürgen

Okay? And the great success story is our PV business unit. It's separated in a separate company because we all know that solar is a high volume, low margin business and there is a roller coaster and it started quite small and is growing exponentially. So at the beginning the story was that we have data loggers as a product and then ship code was in the market and said we will do a house, you're selling all the components and we also need a data logger. So we were trying to sell data logs to Shiko. That was the current CEO from Gundan Instruments Environment Solutions, who is doing a great job. And then Shiko stopped immediately. And then what we do the next thing was well, string monitoring is a topic why we don't measure strings. We can monitor and DataLock everything. So we did start with string equipment and then this was a huge market. We did ship hundreds of channels, thousands of channels to India and all across the world. And then you start to grow, you do combiner boxes, turn key, ship that in containers to the solar boom in Australia and so on. So that team has a great knowhow about logistics and doing big projects and at the beginning it was a megawatt and ten megawatt and nowadays smallest projects are not below 50 MW.

[00:18:56.860] - Jürgen

It's huge. What they are working on is currently 400 MW, the biggest one in Germany. And this is done turnkey with Gantner monitoring, hardware and control and you need a good team who understands the value chain. String monitoring is not the most important thing for the industry all the time. You see it's going down in Europe and globally, but you see small signs again in the US. At the moment, which is funny, because they always rejected string monitoring. And of course, with the central inverters, you don't need that. You have all the data. So the next step was in the evolution that you do data logging and control of the power plants. So we have already all that stuff and now power plant control is one of the key services and we differentiate that clearly that we have open platform and the customer can adapt the control mode. So it's not a black box, it's a platform and we can support that remotely from everywhere in the world. Connectivity is key there and with that we control this one day to log on hundreds of megawatts and collect the same data at the same time.

[00:20:17.190] - Jürgen

And there are more than ten gigawatts out there for the control of Gandhi at any moment of time. So if this is not working, we would have a business problem. So reliable control is key.

[00:20:32.510] - Torsten

So your system includes hardware, right? So it's not only software, it's including.

[00:20:39.060] - Jürgen

Software, it's hardware and software.

[00:20:40.300] - Torsten

It's hardware and software. So you monitor the performance of single small units in the solar park, but also now the full complete solar parks. So what kind of data do you collect and what does your customer do with it and what's the value you create with this monitoring?

[00:21:00.190] - Jürgen

Yes, this answer has to be different per stakeholder group.

[00:21:06.130] - Torsten

Okay, yeah, you're stakeholders.

[00:21:09.630] - Jürgen

Exactly. I will try to group that. So you have different types of phases. Also in the last ten years you have the EPCs in Germany which build and want to sell after two years. So they go for the fast delivery, it has to work. They want to get their PPA signed or feeding tariff and then after two years they want to sell it. They don't care about the data quality, if I may formulate that in that clearness. But the second one who owns who will buy it, he then has to integrate the data again into his platform. So you have already two players, the one is DPC and the second one is the one who is doing a portfolio management.

[00:21:54.820] - Torsten

May I jump in? So the EPC, does he already include your monitoring or he just doesn't care? It's just extra cost?

[00:22:01.720] - Jürgen

No, he is including it. You have two types of companies, a lot of them say I need it, but that was the early days. Nowadays monitoring ordering is defined there and it's below zero 5% of the total system cost. So it's neglectable. But when you want to ramp it up, when you want to sell electricity to the grid, you need that data. So nowadays no longer possible that EPCs can say I don't do monitoring, just do the metering. I have to be clear. Also, when we talk about that, we talk about utility scale counters, not doing the residential and so on. We do a lot of power plants in the air, commercial rooftops for the big mobility brands and all. But then you aggregate that together. Okay, but they always have their platform. Our cloud based solutions offers a service they see instantaneously what is there. So they see if inverter is broken, they have the ticket system, they have document system and all that stuff and track record. Good. And the first two years, if everything is good installed, then it's fine, then you sell it and then other people want to log in and then they want to migrate the data.

[00:23:21.180] - Jürgen

So we are having a separate world, which is more or less the platform management that all the power plants are in there. This is the biggest customer scope and they of course want them to have reporting portfolio management, how is the performance ratio of different power plants and alarming and notifications. So you see if the inverter is broken and so on. Okay. And the third one is the one who really wants to produce kilowatt hours and sell that on the energy trading platforms. They want to optimize the system. And when you want to optimize, you have to understand what is my current performance versus my expected performance. And their modeling and loss analysis comes into the gate.

[00:24:13.240] - Torsten

Excellent. So you have an initial capex and with the software you have a recurring revenue business model or how does it work?

[00:24:26.890] - Jürgen

Yeah, the hardware is based on the requirements from the RFP and all that stuff is combiner boxes and all that stuff. So you have that, they purchase it and they own it. And the cloud based services is a monthly fee that goes that starts on €20 per megawatt. That's roughly the price. And for bigger projects, then you optimize that and then you have additional services for controlling and so on.

[00:24:59.250] - Torsten

Yeah. Excellent. And you mentioned that you're using edge computing. That's a term which is what you hear now often. But what does it actually mean?

[00:25:13.460] - Jürgen

That's correct. It's a buzzword. And since Gantner is doing data logger, since maybe we are the oldest edge computing company in the world, but that's all what I want to see about edge computing. No, it's about data reduction. Because when you have, let's say in each shift we have in Ben Bander, they plotted these multiple plots of 50 megawatt power plants there. And then you have one controller absorbing everything. Maybe you do two controllers because you have a road in the middle of the PV power plants and you want to control it, or you have different owners, so you have two of you, let's call it data loggers in the field. Now you have to acquire all the data at the same time because if it is shifted by a few minutes, it's worthless for analysis later on. But we do that on a 1 second rate everywhere and we are ready also for higher assembling rates already because if you see that in industries PV is really still relying on five minute or 50 minutes average values but customers now use 1 minute but next trend will be 1 second and or faster for a specific time.

[00:26:36.010] - Jürgen

And for that you need data reduction because you can with our technology stream data up to the cloud on ten Hz kilohertz basis for grid quality thing, but you have to reduce that. First of all, if there's an internet connection, you have to buffer it locally, not everywhere in the world you have a stable internet connection or at least wired. So you have to the team is dealing with all the satellite concepts and so on and you also want to trigger activities even when there is no cloud like control. So edge computing is more or less an intelligent device which can work stand alone even when you're not connected to your cloud service. And customers can put functions on that platform so they can apply rules, they can adapt feed in strategies. That's also helpful when we talk about batteries later on. And you can also record data on a trigger base. So for example, when you have tracker movements that you just measure the force and the motor current when the tracker is moving and you don't store data when the tracker is doing nothing. And that can be applied for all different things.

[00:27:57.940] - Jürgen

And out of that you can also do some local logic and that has to be fast and not just on gantner is able to do real time control. So in the high end performance devices you can acquire the data with ten or 100 khz, which is needed for power quality and battery explosion testing. That's too fast what you need for PV, but for PV we use that know how and just cost optimize it for the solar use case.

[00:28:33.710] - Torsten

Another buzzword to edge computing is artificial intelligence. So you have lots of lots of data. Are you providing services in that sector as well? Can you show correlations between wind and I don't know yes. Tracker accuracy or anything like that?

[00:28:57.640] - Jürgen

Yes, and it all starts when you understand one TV device and you scale it up to a utility scale device. So we have also a product line which is doing IV scan testing outdoor that's more or less the DNA of the PV module. When you know how that behaves versus temperance Irradiance, you can already start to model that and there we still work with mechanistic between physical meaningful modeling and not just machine learning. Because when you predict something, you want to hear what happens or what is not okay. But the answer you need also is the why and machine learning is not so far not able to tell you why you have a power drop. The latest trend on machine learning is that you have explainable AI. Now in the field, which is a hot search topics that people try to develop AI, which is explainable what AI is doing. So we test that we do that also intensively because it helps you if you have no data to sanitize and to group but on the PV world you still can do a lot of mechanistic modeling. And there we are in the range of plus minus two and a half percent normalized root means error for the prediction of the power output for any component in the power plant.

[00:30:32.550] - Jürgen

So that's a lot of calculation but you just need the gravity and the module temperature and if it's windy, wind sensor and with that modeling we can really predict the power at any given time and the same also for voltage and current and that also helps us a lot for the fault detection.

[00:30:57.340] - Torsten

So thanks a lot for the explanation just to make it more transparent. So you measure IV, so the current and voltage from the in this case the solar power plant, what else do you measure? It's wind speed. Can you give us an idea of the number of data points.

[00:31:22.390] - Jürgen

That'S in the thousands? So when you have string monitoring inside, you want to measure each current of the string and the voltage of 816 32 strings in parallel depending on the module power. Then you have combiner boxes, combine the boxes are then collecting the current which goes to the inverter. So we measure all the inverter parameters and also the transformers and temperatures and then you want to measure also the grid connection points, power quality, power meters and also the official feeding meters. In big power plants you have several meters, you have incoming energy, outgoing energy and there are power plants is 1050 meters also quite common, especially when the ownership of the power plant has multiple parties. So then we are collecting all the data and the people are looking at that data. Next to that we are collecting of course also environmental parameters, irrigation sensors with reference sales and parameters, wind module temperature, ambient temperature and so on and this is more or less a setup. And you have a weather station for each 20 meg or ten meg depending on the customers requirements. If you have decentral inverters, you just read all the data from the decentral inverters.

[00:32:53.670] - Jürgen

So we do all the protocols there and this brings down thousands of channels per minute in our example here to the data logger and then we have all the data normalized and pre calculated so we can do the enrichment quite easily by the customer. So we have all the reference values and so on. This is in the range of for a 100 megawatt plant, you have roughly several hundred thousand of parameters each minute of time.

[00:33:35.290] - Torsten

Everything is run out by computers. One could naively assume that collecting data is a simple work to be done. What is the complexity in so much data? Could you describe us? Why is it so difficult and why Gunner has so much success in various industries in collecting data?

[00:34:00.490] - Jürgen

It sounds obvious, but it depends always what you want to do with it. If you know that very well, it's an easy task. When we want to measure temperature, you buy a data log or connect the temperature sensor and you have a display. But it depends on what you will do with that display information, because you want to store it over time. So you want to see what was my temperature over the last day? And then you want to see the last year. You need an uptime of 99% whatever percent. So you maybe don't rely on a Windows PC or any Raspberry pi which has a firmware update and after the firmware update it is not running and you have to do the configuration again. So these are typical things from what all consumers know. And when you want to do that for scale and you are responsible that each minute the data is there, you need solid hardware and solid connectivity and you need ups. If there is no electricity, you still want to lock the data. You need different interfaces because each sensor behaves differently. We have all plugins there that you can speak to, all the modbus versions and so on.

[00:35:14.920] - Jürgen

And especially at the beginning of the solar boom, each inverter manufacturer had different protocol nuances. So you have to be flexible. It's not perfectly standardized, even when there is a standard. I think you all know that game. And then you also need the ability that others can have a bi directional communication so that you can send a signal to the date, the logo, which is then sending the signal again back to the inverter and say you are doing power reduction or you're doing active or reactive power and so on. So the control mode comes into that and we combine everything in the same platform, which makes it cheaper for the customer because you don't have a monitoring and the control separately does make sense.

[00:36:01.750] - Torsten

Okay, so reliable data logging is the key here so that it doesn't break down because your customers really depend on high quality data. And then also the usability with bidirectional communication. What's exactly the value you create? Is it more on like initiating preventive maintenance or is it more on optimizing your solar power plant or any other operations? Or is it both? Where do you think is the largest value you create? Primitive maintenance versus optimization.

[00:36:39.410] - Jürgen

It's both or all of them again depending on on the the target customer. Because first of all, you have to have a track record of your asset. When you want to sell your car it's good that the counting number of your distance is reliable. So you rely here that this is traceable and protected. So you need a good data quality which shows how much energy was that asset producing, what are the failures. So it's having data of that asset gives you a higher price when you can sell it in the secondary market. The second thing is you want to optimize the data. So when you're a kilowatt hour driven company you can find out where I'm losing energy and this is also where we provide here the services and you can visualize that have all the alarms there and you can also have the models comparing that. So we can tell you for any given point of time what should come out and what is coming out. And based on the difference you can get your alerts and when you have data over time, you then can do the preventive maintenance discussions which I think is at the beginning of being implemented into the real world.

[00:38:04.500] - Jürgen

There's a lot of bustle around that. But the question is who really runs that in the big fleets? And not just for a prototype or for one of two power plants. And even for that you need high quality data. You cannot correct it over time. And here the big companies are having their own teams working with the data and accessing to our APIs because we have quite frequently the thing that we have the data normalization already perfectly done. So their data scientists can work faster instead of setting up a separate cloud environment. That's the key differentiator. So we are saving manpower.

[00:38:49.260] - Torsten

Okay, now when we look at your customers and maybe you have insights simplify the questions they ask you what you should change in your product or services, what's the key challenge in operating solar power plants? What keeps your customers busy night and day?

[00:39:13.760] - Jürgen

When you enter new areas like Middle East or Australia, which are remote and you are a German EPC or you are a European EPC and you go abroad. It's logistics and to have the right people there and realistic and reliable communication of what do I need? And central point of information is key so that everybody has the same information and the same SLDs and all that stuff and that's number one. So it's the team which is established and a lot of great DPCs and customers have these teams ready and they can really do that on scale. That's really good. When you have that you need somewhere the data so that you can support that from the office. And we don't have a big team flying around. We can do everything from the office with our remote access services and people can prepare everything in the office and then go online and if you have problems, we can log into all the devices and support them. That's very helpful for commissioning. And overall, later on it's the adaptability that yeah, you have phase one and then you want to add phase two. And this has data streams have to come together.

[00:40:33.460] - Jürgen

So this is also very important that people have that ideas. And when you do that three steps, you have to have a flexibility but still robust platform which allows you all that things. And this is very key. And the demand, what people want to do in three to five years, nobody knows, it's not clear. You have to act very quick and on top of that it comes that different countries have different ways, how you control power plants, how you have to monitor power plants. So you have to be very country specific. And it's not that there is a standard way for you, that you have a standard monitoring concept. You have a starting thing and then you adapt it where a lot of more standardization comes in. When the big utilities build up their portfolios, they want to harmonize it everywhere. So they say continental monitoring has to be here. That's the component you have to use. You can select these types of modules, you can select this sub construction component and so on. So there it is. And then you have then power plants built in that pipeline. Then you have a lot of repetition and a lot of good learning effects because people know how it was in the last time.

[00:42:03.340] - Jürgen

That saves the most cost, I think, nowadays.

[00:42:07.090] - Torsten

Okay, so it's cost and one thing, and the other thing is reliability. So do you get insights on how reliable solar modules are these days? Inverters I always said that they break down mostly after ten years. Or in general, what is the weakest component in a solar park nowadays?

[00:42:33.490] - Jürgen

That's the Holy Grail. Everybody wants to have the insights and I'm not able to share all what we see, but I try to share what I think gives us enough insights. The reliability of power plants also depends a little bit of the year or the era they were built. I mean, if you know what was done in Spain, it's the big solar boom now there's a solar boom again and they learn a lot, they do it differently than in the past. Also the big solar boom in the UK, power plants are built up and after one or two years they are sold to another one and they then observe what happens after 510 years. And it's really the important component is always the inverter. And the communication of the inverter is also something which is very important that.

[00:43:29.500] - Torsten

This has to be reliable, the communication of the inverter with the data logger.

[00:43:34.590] - Jürgen

With the data logger, that's an important thing because when the inverter is not able or stable enough to work twenty four seven and send data and you have to reset it. This is something as an example, what are early failures? And we also work with a lot of inverter manufacturers together and optimize and give feedback. And then we have to update the firmwares. We can roll that out to the whole power plant with one click. That's the thing. The second thing is that the components are not able to adapt to the electricity codes. I mean, our plant control is now required everywhere. So if the inverter does not have that ability, you have to exchange it and then you're losing money and it's not controlling fast enough. And this then creates a failure.

[00:44:27.360] - Torsten

So can I jump in? What kind of communication is that? So the inverter receives messages from the grid. For what purpose?

[00:44:39.790] - Jürgen

Okay, I mean, PV is a significant player now in the energy generation and therefore you have also to support the stability of the grid, the 50 Hz has to be stable. So the power plants also smaller power plants have to be controlled. So the utility or the GSO is sending signals. I want to reduce your power, for example, to 80% and then this has to be executed. If this inverter is not able to do that reliable, they are shutting it down. Then you lose everything.

[00:45:20.650] - Torsten

So the grid operator shuts down the solar park because they say you don't follow our rules exactly.

[00:45:28.870] - Jürgen

So that's one idea. Another thing is really that you have to react very quickly and then the ramp rates of inverters have to be controlled and so on. And here we try to bridge the problems from the limitations from the hardware to the grid to not to really adapt to the real use case. But it's also about failure. The question cabling and basic engineering that you do it properly, that the cables are protected from the environment. Also something that a lot of people had to learn, especially when you are going outside of Europe, you understand why in Australia the cables are sealed and protected and sometimes with metal tubes and so on. Termites and all that stuff is really something which you have to take care about failures. And when you have all that basics done, then you start to see, okay, how is my PV device working? And that's also very interesting. That how you get different qualities out there and how the different batches work. Of course you will see that in the meter a little bit, but always have a year too late maybe that you have a problem in degradation. So our job is to really detect that and at least flag there is something running off.

[00:46:59.440] - Jürgen

And you had a lot of interesting insights when you have park technology at the beginning and so on. And a lot of customers are really looking at that from the beginning. So they want to see the Lid effect and so on. And we could really help them to give feedback also from monitoring to their lab based measurements, to the quality operations and so on.

[00:47:26.290] - Torsten

Okay, so you're looking at like the big picture, like a complete strings or the complete solar power park. How large is the need for, let's say, individual solar module measurements? There's mobile labs, so you can do quality control onsite there's inverters per module, which are mostly used in the residential area, but of course, theory could be used in large solar parks as well. So in general, what's the need to have in C two, or let's say a more complicated onsite individual solar module measurement?

[00:48:12.790] - Jürgen

That's a good question, and there is no one sentence answer to that. It depends, again, really on what is your target. I think the big players are all investigating the technologies up front, have their own test labs, have their own test facilities, do the indoor stuff, and for all the components they installed, they have the quality inspections, but that cannot cover how the device behaves over time. And when you know how your technology degrades, if it's degrading, where it is degrading, you can adapt. If you have the problem later on in the field with rearrangement and so on. So if current is degrading, it's different than when you have voltage degrading. If you have to repair stuff like that, that helps, but everybody hopes not to have that problem. So they try to test and at least work a lot with statistics and say, I tested x percent or a fraction of a percent of modules and sample that, and from that I know how this is distributed. So here I think it really makes sense. When you have string monitoring, you can have a lot more insights on the module level versus just decentralized.

[00:49:41.710] - Jürgen

So that's why string monitoring is sometimes added for research projects also on small power plants. But when you have the Ivy scan, you really know what's going on. And I think this is very important that you understand there, because it's not just the maximum PowerPoint area which counts, it's also how is the parallel or serious resistance changing over time. And this is where we give additional insights. And we did prove quite lately and also with a lot of publications, that you can also understand from one module the whole string and conclude a lot of fault behavior from that. So IV scans always make sense from the representative module.

[00:50:33.600] - Torsten

Okay, so you can do IV scans with your product. Can you give us an idea how you do it? Because you only have access to the full string, or do you have a device per module? How do you do that?

[00:50:48.410] - Jürgen

We have a separate product line which is doing IV scans as a solution that's not connected to the power plant, but we can then overlay the data and learn from that. So that's one thing. Another thing is that you can also ingest Ivy scan data to our platform from a third party device and then we can correlate it there and also try to also model it. But the question is always how correct is that? Because you cannot work there with 1 minute average irrigation values. If the sun is moving, you need a higher resolution. So Ivy scans are done in a second. So you have to have the right data acquisition. So we rather add this feature set as a separate system to big power plants or for big customers to offer that service. And then you overlay the data, because when you keep the PV module at MPP between all the IV scans, the results show very clearly that this is very realistic how it is going on. And you can also it's much more simpler to work with the data. So that's the way for the Ivy scan. Also for companies which are doing PV module development research or quality IV scans are always helpful to link that from indoor to outdoor.

[00:52:22.710] - Torsten

Okay, so that's an extra device. You connect to a string. So then you can do different IV scans at different temperatures, different Irradiations level, and do that over time. So you have very lots of data, so you can track degradation or any funny effects very early. Okay. Then there is the non contacting or partially contacting technologies like imaging technologies, electroluminescence and infrared imaging, sometimes done and more and more done with drones. What do you think of that? And maybe can you even integrate this data automatically into your system?

[00:53:07.610] - Jürgen

Yes, I think that topic has big potential and will grow further and further, because it's a different way of getting insights. And I think it's not one or the other, it's when you can combine it. I think you can save money and time. So if you have current of a string and you have a picture of that string and you see how it is behaving with the module temperature or yell and so on, then you can draw fast conclusions. And how we can work on that topic is that people can access our data with APIs structured already. So you can overlay your picture information with the measurement over information and work out here some conclusions so that's possible as of today. And somebody has to work on that enrichment step. I think this is quite in the build up. So we provide you the API at the moment. I think at the end you cannot avoid that somebody has to go there. So it's a quick thing. And if you have no monitoring, it helps you a lot and also see that over time. But also that service is not for free and it's just at the given point of time when you want to run that drone every week.

[00:54:38.590] - Jürgen

Okay, the question is what is the price for that? I have not seen here a business which is profitable, so they rather do it yearly and so on. And then you have to also be clear about the precondition, is it cleaned, what is the angle of incidents and all that stuff. So it's very interesting and I think the challenge is to overlap that with a real measurement and then you have the red inside. What do you think you get?

[00:55:09.790] - Torsten

What do you think, what are other innovations required to optimize solar park operation and maintenance? So we just talked about drones, you have introduced the IV scan methodology. What else would you like to see in the future?

[00:55:31.160] - Jürgen

What I think is key that the data you have, that you utilize that a lot of people are collecting data, but they are not utilizing that enough. First one is the number of data points you have. And that's not something that you can do on your PC. You need really a platform which can handle that. And this is where we work a lot, that you have structured scalable data back ends so that the people can get their data quickly when they need it and out of that data you can do much more analysis. So I think it's underutilized. And the second thing is what I would do before jumping into any machine learning era is let's call it the era of metadata. I think the efficient combination of data, measurement data, time series data with the meta information, that has to be solved. Because I tried to explain that if you have a time series of the inverter performance from yesterday, that's time series data, you can plot that, but you have to normalize it. So you need the nominal power of the inverter, you need the brand, you need the serial number and all that stuff.

[00:56:56.500] - Jürgen

And then you can access that quickly with one API call. Then you can start to analyze more efficiently. And a lot of people spend a lot of their lifetime to combine that data, build it up and then exactly. Can I have it for another version? Yes, go back to square one and then do it again. And this is where our next improvements will be. We have already all the metadata from the components inside and that makes it easy for the people to quickly do their conclusion and do basic analysis to solve the most important things on a daily work. So I think that's quite important. And if you do it right at the beginning, you save time. That's something what I also see a lot. Don't wait to the set up correctly because you need it over decades and every person who will use the data can use it. And if you have the wrong metadata connected, all your work is not helpful, it's not true. So metadata I think is key to combine that.

[00:58:10.460] - Torsten

OK, so if I summarize correctly, correct me if I'm wrong. So number one is use the existing data more wisely, work with it. And the second is add the metadata like serial number product generation, et cetera. So add more information and then again, I guess back to topic one, work with it and work with it more wisely so you don't need, let's say, hardware innovations. It's more like work with the data, create more data and work with it.

[00:58:45.190] - Jürgen

Yeah, and use what is available, I mean use it. We collect really everything what the inverter tells us, even when it's not required because we know later on the customer will ask for it. So it's really collect the data. You don't have to store it forever. But with the experience, we know that a lot of, for example, the temperature is important. At the beginning nobody wants to have it, but now when you start to do aging and things, you want to see the temperature profile of all the devices and quickly you can find out, oh yeah, you forget the shade cover of that inverter in Australia that's worth it. Simple example.

[00:59:30.190] - Torsten

Yeah, let's jump to storage. Obviously, as we have more and more renewable energy in the grid, we need storage to compensate for phases where there's less sun or no wind or even worse in combination. So you mentioned that you also have solutions for the storage battery sections. Just briefly, what is the key challenge here? What's the industry now keeping busy most of the time?

[01:00:04.840] - Jürgen

That's a very good question and I think storage will or has the potential to be a similar even why not a bigger success story than solar. So that's what I think you saw that the market behavior at the moment has a lot of similarities to the PV gold rush when we started 15 years ago, something like that. This is what I see. I think batteries have one more or two more additional complexities. First of all, the business model will change over lifetime and the second one is what do I get? So here, monitoring and quality control is even more important because you have to keep your device at the right temperature, at the right environmental conditions to not lose any warranty claims and all that stuff. So from the delivery, that's important that this is monitored. So this is where we support. Then you install it and you try to run your service. That could, you know, there are more than 23 different services are out there for batteries, different on the things you can have energy Arbitrage, spinning, reserve frequency regulation and black start and all that stuff. But you could also do just peak shaving or backup power or they had the Triads in the UK where a lot of business was done if you take the peak shaving example.

[01:01:46.030] - Jürgen

So we want to shift the peak from PV to the evening, which is more or less now the number one use case in California or in the southeast of the US. Then you do that and you try to collect the energy from the southwest in Arizona and Nevada and provided to california when there's sunrise and sunset and when you have changing energy prices, electricity prices, people could say well, I earn more money when I do frequency control, so let's do frequency control with my asset. And then suddenly you have to be able to control the batteries differently to switch the mount. And this is something that is new for all the players and this is where we really adapt on that. You could run different control mode methods at the same time and for the different assets, because the investors see different opportunities. And I think it's not clear what the batteries will do as a service in the next years. The market is changing very fast, so you have to adapt. And when you make that decision, you also need some history, how many cycles I have done, how is the aging of my battery.

[01:03:07.990] - Jürgen

So there is also a lot of battery analytics done. And here we are. The industry as it's own is learning, is developing skills and a very fast driver is the mobility section where they really try to understand everything.

[01:03:30.710] - Torsten

Excellent. Hey, let's come to an end. One final big question. And when you look at the outside monitoring, you've got a long many years of experience in the solar. You've seen the big rise in the regional drops in the application of solar and now there's a big turn up. Again you see the operation, day to day operation of solar parks. What is required from your point of view to take solar to the next level, including storage, including wind. So that's used more often, more quickly, more globally. What is required to make that possible?

[01:04:28.090] - Jürgen

Big question, try to answer. It not a big answer or long answer, but I think we all see that the energy system should be decoupled from the gas price. This is what helps us because then it comes out. What is the energy production cost of solar? Wind, and it's the cheapest. So it has a lot of momentum at the moment and it will keep, I mean we will see more than 50% of the energy capacity installed will be solar. That this is growing, momentum cannot be changed. I think that's very good. The question is now how you put that into the market. It has to work reliable. You have to have it like a good concept where you can adapt at different places worldwide and it's scalable quite fast, but you still have to keep the quality. I think this is important and it has to play a significant role in the grid stability. So you meet in the middle, you have the big utilities doing the stable grid and you have the smaller players from renewable energy. So you have to be part of the grid of the future, which will be smart in terms of that it is not centralized anymore.

[01:05:49.990] - Jürgen

I think this is needed that you remove that barriers and this barriers should be removed and I think there is a lot of speed reduction tried by a lot of players because you lose power when you have an established energy system with the big players and when it's decentralized, the market mechanisms work differently. I think this is the big thing. The technology is there, you can always improve it, you can reduce CO2 footprint and all that stuff. All this stuff I think is going on and I think the data on top of that should be available and used for all of the participants in the power plant in the energy segment. That you can really optimize that smart system and data quality, I think you cannot argue about that. If you have good quality, you can have good control.

[01:06:54.510] - Torsten

Excellent. So decentralized energy system with smart control, that's the key point you see, to accelerate the rollout of solar and energy wind production. Excellent. Hey Jorgen, thanks a lot for your insights. Thanks for taking the time and all the best for you and Gantner.

[01:07:21.860] - Jürgen

Okay, thank you. Thank you for the opportunity to share our story and I'm looking forward to meet soon in person again.

[01:07:29.230] - Torsten

Yup, bye.

52 views0 comments

Recent Posts

See All


bottom of page