A PVsyst Report Is Not True by Default ft. Orçun Başlak | Soliverse Ep. 47

Coming up

0:00Orcun Baslak: People lie. I enjoy spotting their lies. The degree they are lying is actually the funniest part because sometimes they go way above what is logically understandable, like crazy lengths to make me believe that something is not that.

Solar can't be planned alone

0:22Peter Pongracz: So in life, you know, you also don't have work and uh life completely separated, right? So now the two of them are really bleeding into each other, and um yeah in solar is the it's becoming the same since 2025, right? So you before you it was enough if you only focused on the solar side, you didn't have to worry about the connection, the flexibility, the grid, and all of this stuff.

0:45But today it's a very different situation, right? So you really need to figure out what to do and zoom out a little bit to think about the picture a little bit more holistically, right?

Grid stability, fifteen years on

0:58Speaker: Well, uh that subject has become a little bit more popular given this uh like battery storage and data centers and the AI consumption, but that topic was always there. I mean, almost like 15 years ago when the first power plants was getting into grid in Turkey, I was in close connection with the ministry and the electric transmission company, and they were discussing like how what is the right way to connect them, like how we should be uh handle the transmission, the scaling, everything.

1:39And grid stability has been a part of my life since back then. It is just more popular now.

1:48Peter Pongracz: Yeah, I mean, you know, it's the same with AI, right? Um, nobody was talking about AI in uh 2022, right? And and ever since ChatGPT, what like to 2.0 or 2.5 came out? That's when people really started to see, like, ah, this this might actually work.

2:04Speaker: I think the breaking point in AI, especially in Turkey, is uh, you know, there's a thing called Turkish coffee. Have you ever drink coffee? And oh, I love it. Yeah, and something stays behind, and in Turkey you close it upside down, and some different uh glyps appear inside, and then a fortune teller looks at it and tells you about your future. There is an app for in App Store that you take photos and people comment on it, etc.

2:35And now they pass this duty to Chat GPT, like they're sending their coffee photos and it becomes commenting on them, and that's how the ChatGPT broke out.

2:48Peter Pongracz: Yeah, I mean, but it's always the most uh simple use cases, right? And like have you tried it yourself? What is it telling about your future?

2:56Speaker: Uh I didn't. I didn't, but I talk about my future with the assistant.

3:01Peter Pongracz: That's good. I I try to do the same as well, but it's uh it's quite uh a never-ending top topic, right? So at least for me, you know, all of these AI tools really feel like they give you more information that your human mind can process. So like you can ask a lot of questions, you get a lot of data back, but you're really hitting your limits of your uh you know, monkey brain of okay, I I I cannot cope with this amount of information, you know.

Where AI actually helps

3:30Speaker: Peter, recently I discovered something very interesting. On the weekends, on sometimes in the middle of the week, I take a ride to the mountains, all those roads, etc. Like I enjoy the weather and I drive fast, I enjoy how the car handles the curves, etc. And while doing that, I figured out that actually there's a talk mode in ChatGPT that you can discuss about stuff while driving.

4:04And in that mode, it gives brief information instead of sending you a wall of text. And the day before, I was talking to him about the fiber cables, their carrying capacity, and then uh losing hair and how hair implantation works, and it was like two hours or something. Then yeah, I I was both driving, enjoying the road, and talking about technology, how things work, and I was like, okay, if this is Helen right now, I'm observing.

4:41Peter Pongracz: Yeah, but that's yeah, but I I guess for you, you also don't have an off-button, right? So you're always thinking about like a little bit of work-related stuff as well. And like for me, the biggest problem is that my best ideas usually come at about midnight or like 3 a.m. when the baby wakes me up. So you know, it it's it's it's it's a difficult place for your mind to rest ever.

5:02Speaker: Uh I I I had that situation years ago. Luckily, my my uh I did my tenure.

5:09Peter Pongracz: Yeah, good for you. Like uh I hope I'm also at the end of this road, but uh we shall see soon, you know.

5:16Speaker: Hopefully, Peter.

5:17Peter Pongracz: And speaking about the the AI uh related stuff, you know, like uh we've been chatting about AI and like how that also impacts solar design, right? And you know, if you speak from uh kind of solarian's perspective, like how do you see that change the the whole like project development process as well?

AI in project development

5:35Speaker: Oh AI, uh with respect to project development, I mean if you talk about the development type of tools like selecting a site and doing a thing uh other than uh consuming electricity, AI is currently a helper and a knowledge worker from my point of view because the search algorithms, hence the name algorithm, are still algorithm, same as we use in PDX.

6:08But AI is changing the way we interact with data, and I have a knowledge base of the company, a huge repository that is not really a software repository, but consists of many tools that access different areas, like it access to our accountant and billing, and it also access to our Google Ads. So it's a huge repository. And if I try to make uh something out of data, I am limited by the graphs or what is telling me.

6:45But if AI comes into the picture, he looks at from a different angle and gives you a different perspective. And I think from a uh development phase, uh development point of view, it is very important because, for example, stock markets. Someone says, like, this company is going up and you buy the shares, and some says this company is going down, then you sell the shares.

7:10But if you had a tool to analyze all the noise and everything and give you a different perspective contrary to what is currently happening, it will be nice. So AI, it's it's in its current form, will be very available to companies who develop because they will be able to use algorithms in a way that we have never used and find alternatives to what we cannot think of.

7:36Peter Pongracz: True. And and I I think from my perspective, where I find the most value is it really thinks differently to myself, and I try to force it to be a little bit colder, so to quantify the things, the opportunities, and all of these, you know, from from business to accounting, as you mentioned, all the different aspects, but sometimes it uncovers things and angles you would never come across with a human mind, right?

8:02Speaker: Exactly, exactly. And I love this. If I challenge it, some modules are uh like obeying you. If you say something wrong and they say, aha, you are right, and then they just pass on. But especially in Fable, he pushes me to the edges, like like he he argues to me, and I am like, I'm not I feel like I'm being treated better because it also not just hands out the information to me, but it forces me to think.

MCP and how PVX builds

8:33Peter Pongracz: That's the most value, right? And and you know, if we if we connect this to the the solar design part, you know, I uh you know there is no secret we are we are working together uh uh at PVX, but the how the AI actually works when it comes to solar design, right? So it's uh the first uh or actually second iteration of the the MCP, but it's actually a really cool tool, right?

8:56Speaker: Yeah, uh I actually uh developed our MCP to apply for intersolar competition for solar something, something. And they didn't include it in the finalist, so we actually lost it. And from my point of view, what they selected was not really an innovation at all. So I'm not really sure if the uh election was very widely, and after the intersolar, I didn't lean on MCP development much, however, we knew people joining our team.

9:33I have seen an increased pattern of using MCP while developing the code, and then if MCP is failing at some point, fixing MCP and then again using it like AutoCAD is like in the middle, the code enters here and does something while writing the code. And debug it and then do something again. If it cannot debug it, develops new tool in the MCP and debugs it again.

10:06So it is I mean, the software development was like a human being an operator and doing stuff, but it is like metrics, you know. Like in the metrics, the roots are coming to you and like hiding you. It is becoming something like this. I like it, but you know, I sometimes find it scary.

10:27Peter Pongracz: I recently watched The Matrix, I think like two months ago, rewatched the first one, and I mean those guys really had a vision who who wrote that movie, and it's even more applicable today than it was back then. You know, it's like 25 years old or something.

10:42Speaker: So tell me, red pill or the blue pill?

10:44Peter Pongracz: Take both and see what happens, you know. But but for for our case, um, it's difficult to say because eventually I think we will end up in a situation where I'm a little bit skeptical and pragmatic about AI, but if the development continues like this, I think it's going to be really a problem for humanity, you know, to stay on top or like to operate uh with our current OS, you know, it's going to be really different.

11:16It's going to redefine what it what it is to be a human.

11:20Speaker: Very likely. Very likely. Currently, AI usage is not uh costing the real cost to the people. It is being subsidized a lot, and there is many VC money going in. But at one point, uh it will be something that we will not be able to take out of our lives. And at that point, we will be AI-based subscribers, and you and having an interface like a computer or internet or any other stuff will not uh matter anymore.

11:52We will just be communicating with something in our mind, I think.

Why solar resists the digital twin

11:57Peter Pongracz: Yeah, that's that's possible, but you know, our luck is that uh solar is still kind of based in the real world, right? There are still the the physical components that you need to pay attention to, and and you know, to I I think there is also quite a lot of still a lot of challenge of translating the physical world into true digital twins, right?

12:18Speaker: There always says because every every physical item is unique in its own nature, like you can't really translate, but you can approximate.

From flat land to 3D terrain

12:27Peter Pongracz: Yeah, that's true, that's true. And I mean, since we are talking about um approximations, you know, back in the day everybody was kind of assuming solar as, you know, or at least the land to be quite flat, right? And then uh flatland really behaves in a almost a two-day 2D uh type of uh manner. And when you guys started PBX, you actually took a more of a 3D approach and thought about more the topography and what that entails, right?

12:55So I think that's kind of the first piece of the puzzle as well.

12:59Speaker: Uh topography is very important in our country because I don't know, we have a tendency to be builders, I mean, Turkish people. So, I mean, if it is not flat, we tend to make it flat. So we don't want to fit to the terrain, we want terrain to fit to us. That's how we operate in the country. So that our tool was developed around this concept, and we figured out that this this issue uh applies to other countries with very high mountains, etc., as well.

13:35And uh that's how we took off because solving solving the soil equation, because many people are attacking this this uh problem from many different angles. For example, TFT manufacturers are trying to manufacture a metal structure that obeys the terrain in the best possible way. They want to avoid the complexity of uh handling the terrain. On the other hand, there are some issues about the structural rigidity, how how solid is your structure is, and as a result, you have to make the land flatter.

14:14And this is the engineering I think we should be talking about in the upcoming years, like TFT or rigid tracker or fixed systems or grading. That's where we are trying to seek the best with what we generate as a tool.

14:30Peter Pongracz: Yeah, but what's interesting to see is uh as well, at least from a European context, you know, you clearly see that there is more trackers coming in and less fixed deals, right? And and of course, the the big, easy uh terrains are gone. So now you really need to figure out like, okay, how can I maximize the kind of more challenging, more difficult terrain that I already have, right?

14:52Speaker: Sure. It's it's it has always been a challenge.

Twenty years of solar design

14:54Peter Pongracz: And when it comes to maybe let's drive this back to the the philosophical question from the beginning. So, like, how do you see kind of the solar design or utility scale market evolving? You know, like you've been in the industry for about 15, 20 years, right? So, like how has it changed from the early days up to today to today?

15:15Speaker: Well, in terms of engineering, I think we have more tools and we spend less time compared to what we spent previously. Uh but I mean the the work output has increased per person a lot. However, in terms of solar deployment, I never thought that it will become this big. Like we knew that it will be very big, but then we're thought about it's getting this big.

15:43And every year there was a speech of like, okay, this year solar is floating out. But one way or another, it always came back. So I am happy about it.

15:54Peter Pongracz: Oh, true. And and I mean, you know, if if you're operating on a global scale, that also means you know you're just trying to shift you know gears and move to different markets of depending on how they interact with the utility scale side of things. And you know, when it comes to like different philosophies and different um ideas of how to build you know like bigger solar parks, what do you think is the the most interesting this year?

Solar alone won't hold the grid

16:22Speaker: Okay, I want to circle back to the first question about the grid integrity. Solar in one location being huge is actually not the right way to design from the grid perspective. Because okay, we have solar, it is very nice, uh, we like it, environment friendly, etc. But what we need more is we need the electricity right now. We are having a conversation online and our computers are running, your refrigerator is running, like your kids are benefiting something that's on power right now.

16:58Like everybody is at is doing something on power. So the most important part is the keeping the grid alive. Like whatever happens, we have to keep the grid alive, and solar only by itself is the wrong solution for this requirement. Solar with the batteries is like 50% of the solution. Having all world run on battery, I'm not sure if this is possible, and I don't think so.

17:26It will be possible in the next 100 years, maybe, maybe more. Uh, but we need first grid to the uh up. So whatever you are designing, let it be solar, solar plus storage, or just storage, or maybe a combination of wind, solar, storage, or whatever it is. We have to design for systems that are available when needed, and they are self-sustaining. That should be the ultimate goal.

17:55And solar is just a small part of this puzzle, and you know what? This this whole puzzle eventually, however it is composed of will be very big, and solar eventually will become a small part of it. But that's a very long, long vision. Like I'm talking about like 40-50 years ahead.

18:16Peter Pongracz: Yeah, I mean, but you you know how these things go, like people are estimating long-time horizons and something breaks through, and then it becomes much faster.

18:24Speaker: Yeah, anything else can pop up, like any different type of PV module or different type of maybe like pocket-sized nuclear generators will be the next trend. So anything that keeps the grid stable will stay.

Gas, regulation and Europe

18:39Peter Pongracz: Yeah, and the the interesting thing is I just read it on the on the news today, you know, that um Germ uh the European Commission uh approved uh Germany's package, you know, to basically like do and build, uh fire up any type of even gas power plant just to be able to you know bridge the gap and you know you don't have solar, you don't have wind. Uh in the winter, especially we are heading into uh European winter, you know, the gas prices are already going up, so it's uh it's a quite uh big and tricky uh problem to solve, right?

19:10Speaker: Well, depends on where you're looking at. I mean, since you're in Europe and you have a liberal market, then yeah, you have a problem to solve. But for example, we had the same issue in Turkey. Uh, our president said like the gas prices will be fixed at this money, so we got them fixed. Problem solved. Like, it depends on where you live. I mean, Europe, you European people will most likely regulate it somehow.

19:41Peter Pongracz: Yeah, we we we like to regulate stuff for sure, and I don't know if that's the correct solution, but is it it is definitely a solution, right?

19:50Speaker: Over-regulating is the solution. There's a guy on YouTube, there's a blonde guy, and he always compares US versus EU. And there was a very funny, funny video I saw yesterday. Uh he shows like uh Elon Musk on the stage talking about the robots, the rockets firing, and all the technological stuff that is happening, and it switches back to EU, and there's this lady or Sulawon Mayor or something, and he says, Right now, we have a very nice regulation for you.

20:25All the cashiers that you have at the supermarkets are now running on solar power. Oh, wait a sec, it's not running, guys. Where is the solar power? And then she puts a diesel generator on it.

Cut and fill: three methods

20:39Peter Pongracz: Yeah, but that's the that's that's the irony of it, you know. And it's kind of kind of true as well. So we've talked about you know solar terrain, grids, and batteries a little bit, but if you look at the the first bigger question of how we actually figure out what to do with the terrain, you know, the cotton field analysis, there are a hundred different ways of doing it, but I guess almost all of them are equally wrong, right?

21:06So it really depends on on your exact use case. And if you take into account even uh uh terrain following trackers, then it's going to be a whole nother uh uh way of approaching it. What's your take on this? Like how how do you think about cut and fill and and and how it should be handled?

21:24Speaker: Yeah, uh, I think there are two sides to this question. The first one is the human perception side, and the second one is what the tools can offer. So I will I'll start with the what the tools can offer. Uh we went really deep into this cut and feel stuff. Currently, we have three different grading methodologies buried in the app, and we are trying to find the right approach to apply to a land.

21:59The one of them is what we call smoothing, that you give a terrain and it gives you a nicer, smoother surface, and it gives you a nice area to build a power plant, which, from my point of view, hits the spot on uh at the first go. However, it sometimes estimates cut and field amounts a little bit higher. And then you switch back to something like pad grading, what is uh what works like you have a whole tracker or fixed track, and then you grade beneath it as a flat area, and then you apply some sort of smooth transitions between the roads and the other uh racks, etc.

22:48It is a very nice and applicable and general way of applying the grading. It is my first take for uh most projects, but it is also not enough. So, some in some cases the people just want to hit the limit of the least amount of cut and feel. And this is where the human factor gets in. Because in a recent conversation with one of our recalls, like he was he was desperate, and what he told me, like, Orchard, I need your help to get the least amount of grading, that complies.

23:20Uh, I told him, like, well, we can estimate it, but it may not be a surface that is actually buildable, and said, I don't care, just make it compatible, eat the least amount of grading. It is the people on the site that has to handle it. And for such cases, we have a solver that shifts the rack up and down within the given review limits and grades just beneath the areas that somehow like hits the uh modules at some degrees or some piles that the reveal amount does not match.

23:58It ends up with a very small amount of grading eventually, but can you build it? Mostly not. It is not really really dealable, especially if the slope is very high. So, what we are trying to do here is I want to simplify and merge all those different approaches and algorithms into one thing that people can understand, and I want to present them with a very easy-to-use slider where they can select like less grading, better looking, or somewhere in the middle, at least buildable.

24:40So if they're based on the human perception, they will be able to just uh interact with the slider, and the algorithm will do whatever needed in the background and provide the user with a very, very nice surface. Well, and this actually is the problem. The definition of nice differs among people, like nice for me is a very nice looking power plant, nice for him is the least amount of captain feel.

25:12Peter Pongracz: Yeah, but I guess the the devil is always in the detail, right? Like you can build the best tool, but people still need to use their better judgment as well to a certain extent, you know, because uh you your example was kind of perfect that you know I don't care if it's buildable, you know, it's somebody else's problem. I just want to be within the permission limits and it's fine.

25:31Speaker: Exactly, exactly. Then you keep fighting with your boss, you keep fighting with your colleagues, they just want to get a talk.

25:39Peter Pongracz: I think it also comes down to a little bit of um two different viewpoints, whether you are um just building the project and you know handing it off and you know, not my problem, okay, besides the contractual obligations. But the on the other hand, if you are building and owning, you are probably paying a little bit more attention to these kind of things, right?

26:01Speaker: In most cases, people don't know what they really want. So you have to show them, you have to uh teach them.

26:09Peter Pongracz: Or guess what what they want and show them that they you know help them find out themselves what they really want, you know, because what they tell you is actually not what they what they want and what they need is two different things, right? That's a difficult, uh difficult task to to complete for sure.

26:28Speaker: Yeah.

Reading soil from satellites

26:29Peter Pongracz: We spoke about the the cut and feel part a little bit, but uh on the other hand, you know, when it comes to topography, there is a lot of question marks hanging in the air that very few people are looking at, right? So if you take, for example, you know what your terrain composition is like, like you know, do I pay the land surveyor immediately, you know, at even the early stages of the sales process and you don't have a higher you know uh capex immediately out of the door, or do I try to figure it out somehow of okay, I'm going to be uh ramming 90% or am I going to be ramming 10% on the on the area?

27:09So that's uh that's another question to answer as well.

27:12Speaker: True. Uh and it it comes to our second part uh that we are trying to handle the complexity of the of the ground. Because sometimes you hand a cut and feel diagram to your client, and when they go to the site, they see that it is all rugs, and cutting that part is not really suitable. So we were trying to figure out a way to find what's the composition of the soil.

27:43And uh we figured out that there are satellites that are providing some sort of Earth data that you can use and analyze and calibrate to what is actually on the field and what is their uh percentage of trustability, like how accurate their data is. And we merge this tool, uh we merge this idea into our tool. So uh using different uh satellite data, different maps, and different other stuff, we are able to uh calculate two very important parameters.

28:20The first one is what we call soil compaction. It actually checks using satellite data if the surface is a very dense, very hard surface, or if it is a very soft surface. Uh when we first discovered this data, we had some idea, but we had a problem uh mapping it. So we slowly got some sites, checked them on the field, use already checked uh fields, and calibrated it over and over again.

28:53And right now we have a very nice algorithm that predicts the uh how hard or easy to drive a pile and how hard how how uh good the resistance once a pile is driven into VLB. And we provide this as a data as a in a map that the user can see when they operate on the domain. And in this in this data, here uh we tell the parts that are hard, that are soft, and there are transitions.

29:28And this is important in two ways currently. Well, this is a preliminary data, it is not it doesn't cover the uh data that you go actually there and check it. But what it provides is if an area is really hard and if there's this uh soft area, this transition zone is very important for you because in the hard areas you have to use a drill and use concrete for some concrete inside, and it increases the capex of the projects.

29:56And in the soft one, if it is not that soft, then you can just casually ram the pile and you will be done with it. This helps the user to estimate the amount of cost that they have to spend on drilling. As a result, they decrease their project costs if they can analyze the terrain first, or they can simply avoid the project because the terrain is too wrecky to handle, or too soft that you have to pour concrete at the bottom of each pile.

Erosion and drainage

30:28So having this cost analysis beforehand is very important, and we also have an erosion tool that we also analyze the soil, if it is clay or something, and the slopes and the uh raining regime. And in this tool, we tell you the Russell erosion rate of each section of the uh soil, and we tell you the soil types. And based on this data, again, you can estimate like if it is rain, combine with a draining system, if it is rain, how it will move, where it will go, will the surface of the terrain will be eroded too much or less?

31:13Water be, I mean, let's say if it's a float surface, and if it is rocky, then the water itself will go down and you'll be totally fine. But if it is clay-like structure, the water doesn't go down, the clay resists it, and the water starts slowly sliding, and it creates corrosion and other risks. And you have to plan you plan your draining system. I mean, in a rocky structure, you don't really need a draining system because the rocks just suck the water out.

31:44But if it is clay, then you need slopes, you have draining systems. You have to do all of them together. So that's how we try to handle this uh complexity and give the user the information they seek so you they can design the power plant in a much, much more efficient and correct way.

32:04Peter Pongracz: So uh we we spoke about you know the Capex and where you move the dirt, but how does that actually impact your full project uh success and visibility as well?

Where the money moves

32:14Speaker: Well, uh whilst holding the grading, there is not one single solution. You can have uh a different type of uh structure for your uh needs, and sometimes you don't want to move the dirt out of the project, or sometimes you don't want to cut, or sometimes you don't want to feel. And based on this variability, we build our tool around the system that enables you to specify like if you want to prioritize cut, it does it so.

32:49If you want to prioritize feel, it also does it so. And if you want to limit the feel of until to some uh degree, it also does that for you. And the idea here is it's a requirement from two parts. The first one is as you talk, is the CapEx. If you have equal amount of grading, then you don't ship the dirt out. Dirt stays in, like you just move it inside, and the uh approximate cost for this is roughly $25 per cubic meters.

33:23However, if you need to ship out, it can be like somewhere from $15 to $40 per cubic meter based on the distance you are shipping. So your costs can increase dramatically. So this is the first part of the equation we are solving. And in some countries, there are environmental requirements that uh forces you not to feel any more than, let's say, 30 centimeters or 50 centimeters, or maybe it is totally blocked.

33:54And in that case, our tool operates within that manner. You give him the limits, and it does its jobs by obeying the limits. So if you want cup, it gives you full cup. If you don't want to feel any more than this, then it doesn't feel, and you comply with what you are required to.

34:15Peter Pongracz: That's quite important. I mean, the the game is getting more and more strict, and of course the margins are squeezed uh more and more. So you you want to get the the the whole task right uh the first time, and you know, most importantly, keeping the the cost down, right?

Engineering as a cost lever

34:30Speaker: Yeah, it all comes down to detailed uh engineering while looking from this perspective. You want to keep the costs down, and keeping the costs down requires you to handle every bit like the thickness of the metal structure you are going to use. For example, tracker manufacturers have like uh thicker metals at the sides and thinner metals inside. They are also trying to optimize.

35:00But this this uh, even though it says okay, does not apply to some countries because some countries treat every single rack as just like one rack operating in the middle of nowhere and prone to uh high winds, you cannot apply it, but yeah, it it happens, uh, and we are trying to use our knowledge to its best to handle such complexities, to drive the costs down while making people to design easier, especially the MCT tool we provide.

Trackers, slope and reveal

35:37Peter Pongracz: And you know, this year and also already last year, trackers are becoming more and more you know mainstream in Europe. You know, before it used to be mostly US, but now it's more in Europe and also you know in Asia as well. And and what's interesting to see is that again, you know, figuring out how to put the trackers or where to place the trackers is a whole nother discipline as well, that that is kind of between decided between the engineer and also the tracker manufacturer, right?

36:06So you know, you've spoken to a lot of uh uh tracker companies as well, and everybody gives a suggestion, but you still need to interpret it to your exact project that you're currently working on.

36:19Speaker: Yeah, every tracker manufacturer uh from an engineering standpoint, they all have slope requirement and reveal requirement. But the way they check those requirements are very different. For example, next tracker is using the distance and distance in the reveal, sorry, uh the difference in the reveal heights between the piles, not as a whole, etc. So they have different solvers.

36:46And while one tracker applies better to one side, it does not apply that good to the other side. So uh in our system, we have this kind of uh checkers and solvers built in. So based on your rule set that you have placed, uh you can check if those trickers are able to fit inside and you can get more installed power out of it or not.

37:14Peter Pongracz: Yeah, and I mean, you know, from speaking to PCs over the last five years, like most EPCs are using like three to five different uh mounting systems, right? So they are just rotating those, they are not going to be going through, you know, selecting out of 50 for one particular project.

37:30Speaker: Well, uh not all EPCs are pushing hard on engineering for such uh approach. I think they are mostly sensitive to pricing of whatever they are using at that point because they are trying to solve the problem by applying the good old grading. But yeah, some some very, very hardcore engineering teams check for this and they don't tie to one supplier at all.

Negative hours and storage

38:00Peter Pongracz: True. And you know, I I guess once you have the trackers figured out, uh now the the the bigger question is flexibility, right? So you want to be producing at the right point in time. So we've seen like a record number of negative hours across the uh definitely in Europe, but uh across the globe as well this year, and you it really taught us that you really want to be producing at the right time, not at uh midday, right?

38:27Speaker: I mean, circling back to our uh first questions, it is again the great stability and batteries are currently the maybe band aid right now, but it is a long time problem solver. They are just slowly starting, like we have the hype and we are solving for the batteries, we are solving the equation for the batteries as well. But yeah, publishing uh generating electricity at the right time is currently only possible with the batteries.

39:00And Peter, you know what? Batteries are also very helpful in stabilizing the grid. It is something we call secondary uh regulation or something. I don't know the exact English word for it, but it creates uh reactive power to stabilize the voltages and the grid to keep us keep everything up and running.

Cybersecurity and the grid

39:23Peter Pongracz: Yeah, balancing the grid and and you know it also ties into a little bit of geopolitics as well, right? That we see more and more like cyber attacks and and you know like uh different different you know unfriendly countries are testing everybody's grid. And I think you know the the the cyber aspect is also going to be uh a bit more important in the next upcoming years.

39:46Speaker: It has always been an important aspect. Uh I remember I was uh fiddling around with data logger maybe 15 years ago that was in the market, and they put their dangerous type of admin password in the comment section of their uh login page. And in in a power plant in Bulgaria, I figured out that if I change the frequency response of the inverters, I can destabilize the grid and shut down other power plugs, and then I can use because the grid operator says like too much power, you have to uh decrease your power output.

40:31And our boss was not happy at the moment back then. I'm talking about 15 years before. And I figured out that I can just destabilize the grid so their protection systems trigger, and our system keeps producing because the total produced on the line goes down, with be with us being the sole producer. And well, I was a kid though back then, like I was the script kid that is somehow able to handle it, and now the things are getting much more sophisticated, not related to solar domain, but I don't know if you are uh watching the news, there is this mythos model people are using in Project Glasswing to check for the issues.

41:18And recently it figured out massive vulnerabilities in Linux kernel, and Linux kernel being compromised means like everything you are running on right now is somewhat somewhat touching a Linux kernel. So it's gonna be dangerous, it's definitely gonna be dangerous.

41:38Peter Pongracz: Yeah, and I mean you know, speaking about the grid, but also like whatever runs critical infrastructure, right? So if you are looking at you know uh the systems that are coordinating air air traffic or you know the yeah, the grid itself, like if you if you cut off the grid, you you saw like panic kicks in probably in like four to five hours, like we we saw it last year in Spain and Portugal.

42:00And I think those things uh should better be avoided, right?

42:04Speaker: Well, uh people have to take this AI thing seriously because if you you can only overcome those if you can use this technology uh in the in the way it is meant to be, instead of just sending your coffee pictures to ChTT.

42:22Peter Pongracz: We talked about quite a lot of topics concerning you know project development and uh the whole EPC game. How do you see that changing today and perhaps for 2027 across the globe?

Data centres and gigawatt parks

42:36Speaker: I think with the introduction of the data centers, this uh solar will just keep going, if this is what you're asking, if I understand your question correctly. And uh the tools of developing and installing and designing power plants uh will again be very important. Uh, because I think we haven't seen the scale we need uh with respect to solar, it is just much more coming, even though it seems slowed, much, much, much more solar will be on the way.

43:17Peter Pongracz: So let's repeat repeat that so much more solar is coming in the future, right? So I mean we see we see the tendency if you look 10 years back. Uh I don't know what the the average size of uh uh a mid-sized project was, but you know, as we are talking to different people across the globe, now we are talking about gigawatt scale solar parks in one site, which is crazy.

43:42And I think it was probably unimaginable five years ago.

43:45Speaker: Gigawatt scale. Well, we had one gigawatt power plant like in 2017 or 18, I think. Like we we started developing it as a country as a big called Ica power plant. And I think we will be seeing much more gigascale harpness. And maybe if the solar module efficiencies approach something like 35 to 40 percent commercially, then we can easily go to the terawatt scale, given that we can utilize the deserts or unused, unexplored areas of Kazakhstan or such countries with vast amounts of land, nobody doing.

44:29Peter Pongracz: But then it becomes a question of transmission, right?

44:32Speaker: Yeah, it will be. It will be. And maybe we should install like 1 million volt uh transmission lines all around the world to transfer power.

44:42Peter Pongracz: Yeah, I mean there are some projects out there that are uh you know trying to do something like that, but I think it's yeah, the the way the world is uh today, it's going to be I think it's going to be more and more difficult between countries and continents to cooperate on like large-scale infrastructure projects, right?

45:01Speaker: They have to eventually. They have to, because they uh collaborated on oil, like uh, for example, natural gas. I mean, in Turkey we have natural gas from Russia, from Ukraine, from uh Azerbaijan. Like all the countries are bringing their natural gas and we are piping it to Europe. And Europe had to, we had to, they had to, they have a product, and Europe wants it, so we became the market.

45:28And eventually solar has become uh one of those items, but eventually electricity again will be open like this. Currently, we have very small cross-border transactions, not huge, but yeah, there will be like I think one big like one million volt scale that handles transmission from a huge uh distance.

45:53Peter Pongracz: Yeah, I mean that that would be the dream, right? If you can transport that energy that you produce, because uh I've been talking to a lot of people in the industry, and uh everybody's theory is more or less the same that with if you can get to like 90-95% uh renewables with the current setup, and then probably the last five percent will have to be filled in by green hydrogen or something like that, and and uh like that's going to be exponentially more difficult to fill you know the last five percent than the the the 50% beforehand that right?

The ceiling on yield accuracy

46:26It might be. So if we think about the the solar engineering part as a discipline, you know, there is uh yield, which is probably one of the most critical components. How is that impacting you know both the design and also the project viability of things?

46:43Speaker: Yeah, you took me back to my first days that I downloaded software called PV Syst, and it has one million buttons that you have to understand and sliders to play around to get a yield analysis. And currently, uh yield analysis has always been an area of interest for me. Uh, and I really like the tool PV Sys, and we are still using it in our uh consultancy company, and we are also using it in our uh PVX software.

47:17But this yield analysis domain is first of all, a tool can approximate or at least try to approximate for many different cases with experience of many years. But first, you have a variability of the weather. I mean, this year it is sunny, next year it will not be that much sunny. So there is a data called T MY, typical meteorological year. That is what we say, like after like many years, this is where the data will converge, and we try to estimate based on that.

47:59But what happens on the field is most of the time different because irradiation varies by nature, and this this variability is between three to six percent for most places, like three every expect a variability of three to six percent, and after this variability, you have a tool that handles the yield, and I think there are no companies that say like you miscalculated the yield, like everybody somewhat calculates the yield, and they end up with very close results.

48:39We have a very nice open source library called PV lib that has all the different irradiances, the algorithms, the calculations, the losses that you can play around. And we have the PV Syst itself is a proven software in many domains and it's currently leading the market. And it is uh relatively inextensive software for anybody to play around. And there are also other companies that are trying to compete in this domain, saying that they are more accurate compared to what's on the market.

49:18Can you prove them right? No. Can you prove them wrong? Again, no. So it is an unknown stuff, and there are other technologies like rate tracing. I mean, PVCs, as far as I know, don't do ray tracing. And they say like we do the rate tracing then and we calculate like 0.5% more accurately. Um, okay, but we have 5-6% variability in the in the uh meteorological data.

49:48So how can I be sure that your algorithm is actually doing? I mean, let's think about the cables. Like you lay the cables, you say like I have 2% loss, maybe the cables are longer than what you already know, and your actual cabling losses are 2.5%. So how how can your irradiation algorithm handle this properly? So eventually it is a tool just that approximates.

50:17What I so I cannot say this tool is better and this tool is worse, but PV Sys has their name, and I tend to stick with it. Not because they are the best tool, because from my point of view, as a user, and and this is this is I think the the most problematic part about PV Sys, you have to know the tool. A PVC's report is not true by default, because there are knobs that you have to play around to get them right, so the tool can approximate correctly.

50:50And there are many tools that you have, many details that you have to play around. Just because it is a PV Sys report does not mean that that report is correct, it is the user. So if you at least take the complexity of entering this data and let a known engine, as let's say TV Sys is an engine, run it and you get a nice result that is legible, then I think for the most part you are fine.

51:21Peter Pongracz: Makes sense. And and you know, there are there are many different projects, you know, trying to uh offer better solutions to PV Sys, such as uh plant predicts from Terabase, you know, and other tools out there on the market. But as you as you said at the beginning, it's difficult to prove them right and difficult to prove them wrong as well.

Yield as a marketing domain

51:40Speaker: This yield analysis domain is a marketing domain. It is all about how much money you can pour into thinking that your product is superior. If you can overcome this, then you can be a challenger to PV Syst, not because your tool is much more accurate and approved by three to five universities, but because you have the money to make them believe it is.

52:06Peter Pongracz: Yeah, indeed. But you know, the the keyword here with PV Syst is bankability, right? So uh just like uh with other tools, like if this is the the kind of use to workflow of how banks are used to underwriting, and and you know, it's difficult to change, right? Because by nature banks are risk averse, so it's it's going to be quite uh challenging to you know promote or even present a new uh model of simulating the yield.

How a PAN file fakes bankability

52:34Speaker: Yeah, Peter, but bankability, again, this comes to the thing that about people operating it. Uh in our engineering consultancy company, we had a solar module manufacturer. And since T V Sys is a so detailed tool that it has incidence angle modifier section that you can play around for the solar modules, from a technical perspective, it is a very great tool.

52:59Like it it calculates the angle of the sun that is uh irradiation coming, and it calculates how much of it actually penetrates it at what angle. Like if the sun is setting, then let's say this is your module, it comes like 90% of it just goes back and only 10% goes in. Very nice technical knowledge. And I am not a good solar module manufacturer, and I play with my pen file and I make it flat.

53:27So I say even though the irradiation comes at 90 degrees, all of it goes down. A very small change in the algorithm, and bada boom, my irradiation, my uh production is higher compared to the other uh solar module companies. And you know what? No finance guy can understand this. It is not in the report. You have to see that something is not right with it, understand it.

53:56So it is again the user, like a user can make you believe that this project is actually uh bankable. Well, it's not, and you get the PVC report for it.

54:08Peter Pongracz: If you think about your um solar engineering days and and and uh the owner's engineering side of things, what's the craziest example you've ever seen on a project?

The craziest thing he's seen

54:18Speaker: People lie. I like them when they lie. I I enjoy spotting their lies, but the degree they are lying is actually the uh the most the funniest part because sometimes they go way above what is logically understandable, like like crazy lengths to make me believe that something is not that. And uh, I don't know, maybe he's or she's a three-year-old, sorry, uh years of experience engineer, and I'm sitting here with 20 years.

54:53I know, I know how the internals of PV sys and pv leak work, I know the algorithms behind, I know how the work is done, and they are relentlessly trying to convince me that this is not right. And I say, okay, I wish you a good day.

55:13Peter Pongracz: And is is this uh from a sales perspective or also like a compliance and checkup perspective?

55:19Speaker: Yeah, well, mostly mostly for the compliance. Uh in the in the beginning of the uh solar procurement and EPC contract, there's something called PRs and ratios and uh warranties that EPC gives to the investor. And they are very important numbers. You they because if they don't obey it, eventually they have to pay for what's being lost. So they have to make up for their losses.

55:43And calculating them is a very delicate business that you have to handle properly with IUC standards, the algorithms, etc. At that part, since the money is involved, they can do many tricks. Many tricks. But if you asked me, if you asked me, okay, I remember what is the funniest thing that I have ever encountered. Uh, we have a software for checking the duplicate EL images, you know, electro limit images of the solar modules like the cracks and so on.

56:14We have software for checking it because in one company we spotted that after 6,000 modules, they start using the same EL image with their serial number. So we developed this tool, and we are using it occasionally in all of our products, uh, sorry, uh, inspections. And in one inspection, like we were inspecting, I don't know, 50,000 solar modules. And the person that is in our team told me, like, uh, or to me, we have only have four unique images.

56:49And I was like, What?

56:51Peter Pongracz: That are rotating, yeah, yeah, yeah.

56:53Speaker: Four images rotating for 50,000 images. And I was like, You sure? And she was like, Yeah, come on, let's have a look. And one of them has a spot, like a little crack, and every fifth, uh, sorry, fourth image has that crap, but the serial numbers are different. And we called the uh we okay, this this is the funny part. We we called the uh factory and said, Look, this you have a problem.

57:20I mean, this is this is not obvious. And the guy told me, You don't know anything about solar. We are so so so good that all the EL images of our solar modules are the same. I was like, wow, this is the biggest lie I ever heard in my life. You know what? The factory got closed.

57:45Peter Pongracz: Yeah, I mean, so so it should have, but you know, like a good attempt, like it's it's clear, you know, like if you put so much effort into faking it, like you could probably put the same effort into fixing it as well, right?

57:56Speaker: True. Like that that guy was, I don't know, crazy, huge liar, great liar. If you want to lie, you have to lie big.

58:06Peter Pongracz: So, Ocho, we talked about a lot of things. We talked about cut and field, we talked about soil compaction, water raise, uh yield simulation, uh and and everything under the sun. But I I guess all of these things culminate in one single thing, right? So you need to be able to present your project and what you have actually designed. So uh how how do you approach that?

58:28What is your experience with this?

Presenting the project

58:30Speaker: Yeah, uh, for presenting your project, uh we have two ways. One of them is the WebVieve part, which is something just you click a button and your design ships to our cloud servers, and you display it in a nice JavaScript page that you can save it to your uh clients or superiors so they can play around in their mobile devices, like it opens in an iPhone or Android, and they can see it around.

59:02But we have our assessment tool, and it is the data you know and the data you don't know came in together, so it generates a 50 40-50 page of assessment report, and it says, Look, your project is here, you have those power plants nearby, you have those transmission lines, overhead lines. It gives you details and location of those overhead lines, the power plants, who owns the power plant, what are their powers.

59:41It tells you about your soil structure, like if it is clay or anything, like the hardness map. If you have applied uh grading, then it shows you the details of the grading you have done in a nice and understandable way for anyone. Tells you about the cut and field volumes, uh, where this volume is happening, like in hard surface, short surface. And then it goes into the electrical details.

1:00:09Since the software knows about your electrical design, your wires, diameters, IC standards, and everything, it slowly tells you more about your project. And this is this is one of the most important parts. Uh, as I told you with the PV Sys, you have PV Syst, but you also need a good operator. Here we are trying to remove this uh complexity of uh using PV Syst.

1:00:41We have our uh data here, like all the power plant, etc. The albedo, for example, we have the albedo data from the satellites, and we tell you what data we have about your power plant and what data is suitable and it should go into your yield analysis. And then we ship this data into the PD Sys engine, and this engine generates uh values, and then we use those values to generate our report, the graphs, etc.

1:01:14And then we, if there's battery storage, we also incorporate better storage into the algorithm and get data out of it. And then we also ask you for the financials. Then we correlate this relation with the financials, the suitability of the design, we give them a turnout analysis, uh, DCAC ratio optimization, everything. And all this information you have seen up until now goes through AI, and AI creates you a very nice executive summary that says look, here you have a flooding risk because there is a river nearby, you have erosion risk, you have too much cut and feel compared to general uh expectations of a solar power plant.

1:02:05You are expected to spend this amount of money, and given your financial structure and this project, your return happens in six years. Your NPV is slow, your NPV is fast, and it gives you this nice executive view so you can understand it. And if you want detail, you can just go to its page and check the details. It is not just a system-generated yield page that gives you lots of data that you have to make sense of.

1:02:34It is something that a human prepares for a human.

Where to find Orçun

1:02:39Peter Pongracz: Okay, cool. But I guess if people want to see this, we will leave a link in the show notes below. What's the best way to reach out to either PVX or yourself?

1:02:47Speaker: In our website, we have a uh meeting tool that you just pick the day and the time and send the meeting request, it will be automatically set in your calendars. Uh, because the software, well, with all the features, it is really nice to have. But what's most important is to have people succeed in what we are doing. So uh we want them to reach us. So reaching us is really easy.

1:03:14You just go to our website and we are done for you.

1:03:17Peter Pongracz: And uh, Ochun, thank you very much for sharing your time with me today. It's it's it's been our second discussion, always a pleasure. And uh, let's do this again soon.

1:03:25Speaker: It's my pleasure, Peter. Thank you for having me on your podcast and talking more about solar and how the technology changes our lives. It's my pleasure to be here. Thank you.