Lightsource bp’s Data Strategy to Drive Performance & Innovation in Solar

The transcription of the video is below.

Introduction

Kerim: All right. Hi, everyone. This is Kerim, Kerim Baran, with SolarAcademy. Today, I’m here with Armando Solis and Jimena Martinez from Lightsource bp U.S. Operations. They are in the engineering function of the Leading Development Company. And we’re going to talk about what solar industry thought leaders are doing in their strategic approach to drive performance and innovation by leveraging and standardizing data in their operations. 

As many solar industry players know, there’s data all over the development process, starting at the very early stages of development, going into selecting sites, starting construction, building the assets, and then operating the assets.

Lightsource bp as a major developer that’s developed north of 10 gigawatts worldwide, managing north of 3 gigawatts of assets just alone here in the U.S. market. They’re dealing with a lot of data, obviously. They’re constantly looking to optimize their operations.

Lightsource bp’s reasons for starting a data initiative, data driven operational improvements

So with that, Armando, as the VP of Engineering for US, you guys have started a data initiative internally, to optimize your use of data and obviously, you guys are using many different tools and SaaS platforms.

So can you tell us a little bit about like before this initiative and after this initiative, what was the reason you guys started this initiative and with Jimena leading it, how did things evolve, and what were the problems you were facing? What are you guys doing now that is a change versus the olden days? And then we’ll dive deeper into all the good things you’re doing.

Armando: Sure. So I mean in the industry, I think everybody has a really good understanding of collecting data from an operational facility, everything from meter data, MET data that we use to characterize performance, but there’s a lot of other data that we have in our projects that we don’t necessarily use or leverage. Over the course of these first couple of years of operations here in the US, we’ve started to see how much more value that useful data can bring us.

And we’ve looked at some third party vendors, suppliers all across the project life cycle. So today, we’re going to focus a lot more on the operational side, but we do see a lot of value across the entire fleet, and where our business starts to where we start the operational portion.

One of the things that I wanted to help exemplify, especially through the work that Jimena has brought up, is leveraging data to help us optimize, as you just said before, performance, and really make these projects generate more energy. 

Now, that’s not to say harnessing more power from the sun. It’s really being efficient and effective in the operations and the maintenance services that we do. So there’s a lot of this data that we can use to leverage personnel time, etc. But it all takes effort to put it into one place that makes it usable.

Kerim: Got it. And with that, you guys, I guess, gave the charge to Jimena to start analyzing or looking at all the different buckets of data internally, and then come up with some methods and approaches internally, and to, I guess, standardize and optimize which data to prioritize and how to collect it all and how to make sense of it all.

Armando: Yeah.

Lightsource bp’s internal data initiative – STORM: Site Tool for Optimized Resource Management

Kerim: With that, Jimena, maybe you can tell us a little bit about this initiative you’ve been leading internally and some of the insights you’ve encountered in your efforts to make sense of this data.

Jimena: Yeah, so we look at very different types of data, all the time. You know, we have SCADA data that gives us meter power, inverter power, MET stations like Armando said. So that gives us insight into the operations of the plant. But we also have data, for an example is like, who is the EOR? Who is the mechanical EOR, what type of –

Kerim: What’s EOR?

Jimena: Engineer of Record.

Kerim: Yep, Engineer of Record.

Jimena: So there’s a lot more data that doesn’t come from the SCADA that we’re trying to capture. And earlier this year, Armando tasked me with finding a better way to use this data and a better way to capture this data. That’s where STORM came to light. 

So myself and Juliette and Connor, my two interns for the summer, started thinking about all the different resources that we have for each site, and you know, in our team, different people that have different inputs into this data and how we capture it all. So this tool is called STORM. It stands for Site Tool for Optimized Resource Management.

Essentially, it starts with a form, with an input form that goes to the project engineers. So the project engineers are responsible for building the projects to spec and guiding all the technical discussions around that. So they’re the ones who give us what inverter type it is, what module type it is, what tracker, what SCADA system.

And like I said, who was the electrical EOR, who was the civil engineer, who was, you know, like all these different points that we get even before we have data for the plant.

So they –

Kerim: Is this collected even before going live during the construction process?

Jimena: Yeah. Because, for example, we get different sets of as-builts or drawings for the site. As-builts are at the time of substantial completion, once the EPC is done, you get the last set of electrical drawings. But then, before that, there’s the 30%, 60%, 90%, but we always know, early on, what is the equipment on site, what fuses we’re going to use. And sometimes that changes. So STORM starts with an input form where they fill out all the information. 

Then the second layer would be the project manager or the construction manager. They add different things that came along the way. And they’re the ones who actually have to input the last set of documentations, and verify that we have the weather stations working, and that we have the MET layouts, and that they’re kind of like supporting roles, but they’re two different types of data inputs.

After that the next person responsible would be the O&M manager. So the O&M manager inputs anywhere from, is there an O&M building? Do we have sheep on site? Do we have video cameras? So there’s like different people who are responsible for different data at different times.

So what we did at STORM is just collected it all, and we’re able to use the technical data, the admin data so we know what portfolio the site belongs to.

Then we also have the resources, and the resources can be internal or external. It can be anywhere from an electrical as-built, an inverter data sheet, but it can also be a third party SaaS, for like a drone flyover, or a SCADA platform, or a soiling station platform, I think there’s Fracsun Soiling Station. 

So we have links to all of that through a single tool called STORM, and it’s STORM because it’s a little bit funny, because it’s a STORM of documents that we have to deal with sometimes. So it’s really easy to just have them kind of consolidated in a single place. 

And then the last person who’s ultimately in charge of maintaining this data is the asset manager. So, yeah that’s STORM.

Creating a Digital Twin of a Solar Plant

Kerim: And this is all in the purpose of creating essentially, a digital twin of the asset on the ground, with full on, almost like a health records for humans. But it would be the health records of a solar asset or a power plant, essentially. Is that a fair way to –

Jimena: Yeah. I’m personally really excited about where the industry is going in terms of having a digital twin. I think we’re able to get to that right now with drone technology surveying the site. And we get what’s called the digital twin. It essentially is a live map of the site, but as we go on, we’re able to feed more information into the digital twin. Right? 

So a digital twin, essentially what it is, is a digital mock up of the site that results in a single module, has a place, and then it has a history. So if that module is replaced, essentially, in the digital twin, you can see that. That module has been replaced. Or an inverter faults, or all these issues are kind of layered on top of the digital twin. And ultimately, we’ll be able to have a really accurate history of the site with all of its components, with recurring information that just updates constantly, because these sites are, you know, meant to live for 25-35 years. And there’s a lot that happens between them. So it’s really important that we start keeping record.

And then also, with the digital twin, we can go as far back as development. We can utilize the surveys and the hydrology assessment, and see how those have evolved over time.

So we’re trying to put together the full circle, the life of the asset within a digital twin. So I’m really excited.

How Lightsource bp prioritizes layers of data to collect and use

Kerim: When you create this digital twin, I mean, you guys are probably using some industry-leading tools like PV case, maybe other design tools, some O&M software. Do you create your own, or do you collect your own raw data layer? Do you create that? And then like, maintain it on that internally? Or do you connect all the data between the different SaaS tools? What are some such tools you’re using, besides your existing SCADA systems or other systems?

It must get quite complicated. How do you prioritize which data and what to collect, what to prioritize, what to organize? I’m assuming right now you guys are focusing on more the live asset piece and like, what are your future plans as you evolve this more data centric management of these power plants?

Armando: So maybe I can take that question. There are quite a few softwares that the team is using cross functionally. From the beginning of the life cycle, we tend to use what we would consider rapid design tools. Or, yeah, I mean, that’s really the term that is considered.

Kerim: What are some examples of those rapid design tools?

Armando: Yeah. PVcase, PlantPredict, RatedPower, which I believe they changed their name to something different obviously, autocad. But there are a lot of tools and cross-functional platforms that we can use to combine just engineering software and those. But then there’s leverage of GIS platforms that we have with a fairly robust GIS team that we have internally. 

So from the very beginning, we start putting all of that data into a GIS platform. Eventually, as Jimena said, we’re trying to collect all of this data. There are so many different sources from O&M logs to equipment supply, the rosters, all of these things into one place. So it’s that unification that is the ultimate vision, as Jimena said, a digital twin that has all of this metadata and attributes tied to it. So the platform that we’re probably navigating toward and maybe I don’t want to spill too much of what our secret sauce is, at the same time, I think this is what the industry needs is to help standardize is One Map. Right? 

One Map is a very large software platform that is used not just in the solar industry, but in a lot of other different types of construction and other energy, and so many other platforms. So leveraging that to help combine all of these relevant items is where we are looking into, whether it’s pulling drone information from layers that we collect, whether it’s internally or through a third party or drawing and you know, survey data that we’re entering in.

The second part of the next phase of it is tying all of these attributes to all of these different components and leveraging that information as we get into that operational phase.

Kerim: And that can also take you – oh, go ahead – Jimena, please. 

Jimena: No. I was just going to add that, for example, I think Lightsource, and maybe other companies that operate internationally, we face the challenge that in Australia, they might have a provider, and then in the US, we have a different provider, and even in the US alone, we have different providers for different things. We have different O&Ms, we have different EPCs. So we need to work to build our systems, to be able to have inputs from all these different providers worldwide.

So we need a way to have our data centralized and standardized for our business purposes. But we also need to be able to be agnostic to who provides the services because we’re not experts in every single platform. We’re not software developers. We’re not trying to beat people in their fields, or be the best at everything. But we do need to be able to ingest all that information and consolidate it in a way that it informs the business as a whole, regardless of where the asset is being built or operated.

Kerim: And this makes a lot of sense as the industry is growing. And you know, and everyone’s operations are getting so much more complicated and sophisticated. It’s great that leaders of the industry are thinking about standardizing these approaches to data collection and management.

What is Lightsource bp’s North Star when managing data?

I’m assuming you guys have probably a dozen plus different software tools that you use in your processes, and it must be quite a task to prioritize which layer to do first, and then the next layer, and also scope-wise like, okay, operating the real asset. But then, maybe, like focusing on the earlier stages of development as well.

So when you think about like the North Star of your approach here, what do you think of like as being the ultimate goal of all this data activity, if you will?

Jimena: Well, I think my North Star is always, you know why we do this, and that for me, it’s provide clean, reliable, renewable power, and that performs, right? Because you can’t have a reliable system if you have an underperforming plant.

So I think it all works together and performance of the plant. These are essentially living, breathing things, you know? If an inverter goes down, the plant goes down, performance goes down, so I think that that’s our North Star, just having good reliable power on the grid and having good performing sites overall. And that comes with having the right information and having the right tools and having the right data to diagnose the issues. And it’s all within the same realm.

Kerim: Got it. And – 

Armando: I was going to say, if I were to give my North Star in terms of  what this means, there’s an element around our industry that, and maybe it’s certain feelings, or how people see what solar or what renewables mean is, it’s always lean, cheap, and in some cases, even a race to the bottom. Right?

Kerim: Yeah.

Armando: Our industry is very competitive amongst, you know, our peers and our other developers, and then also our contractors, all of our business partners, our suppliers, etc. There always seems to be this notion of, at least in the U.S. market, a race to what is the cheapest form of delivery?

And by leveraging data, by looking at what all of this data means it is going to help make a competitive edge on how we become more efficient, and a lot of it means, like how we spend our time to find more of these opportunities to deliver better assets. We deal with markets where solar may not be the most favored. So we have to use data in those phases to identify better solutions. 

When it comes to operations, we’ve seen a lot of the most recent media coverage of the storms that have decimated project sites. So how do we use data to help prepare those sites, for those types of incidents, whether it’s hail, wind, or otherwise? And then how we make equipment selection based on that location, all of those inputs of data, we can use and become more, not just a better competitor in the market, to win more projects. But also, drive the market towards better resiliency than just cutting costs.

Jimena: And a big one that supports that, I think it’s predictive maintenance. And that’s where we’re trying to get at. So, for example, like there’s multiple inverter OEMs out there, and they all have their thing. But it’s how you prepare for the events that are going to cause downtime. So you can get to a point where, if you analyze the data long enough and deep enough, you can understand that, for example, close to the summer in Texas, certain OEMs have fan failures. And if you don’t have the right piece or the right fan within your spare parts, then you have a de-rated inverter for as long as it takes for the part to arrive.

So these are the type of insights that we’re driving in terms of what caused loss of production? And how we can prevent it in the future, so that we have extra amount of fans right before the summer, and we mitigate that kind of lost production. 

So we’re really trying to get to that point where we’re trying to predict and be ready for those instances.

Kerim: That makes a lot of sense. 

How to standardize data processes across EPC partners?

And another challenge must be that you guys are probably working with dozens of different EPC partners because you don’t have the workforce to develop all your plants, I would assume. And then how easy is it to – I don’t want to use the word enforce, but encourage those EPC partners to think about data in the same way you guys are, and to input all the pieces of data that is necessary during the construction, post construction. 

I don’t know if you necessarily use their workforce for O&M functions, post, go live. How do you approach that problem?

Armando: So I can certainly say, that’s probably one of the bigger challenges. And it’s not to say that an EPC is going to be that challenge. We hire our EPC partners to deliver. That is their bread and butter. That’s what they do. Our business is to bring these projects forward, so it becomes very hard for us to navigate the line of being too prescriptive, but also asking for the information that we want. 

So to some of the points that Jimena made earlier around being agnostic, it doesn’t just evolve to a global problem of how we navigate different markets. It also becomes a smaller problem here when it comes to navigating different contractors. So really, we have to look in and see how we can ask for all of the same information in a simple way that we can extract all of it and put it into our tools and systems. 

Obviously, the objective would be for all of them to use that same platform. And I think we can get there at some point, but with what we have now, that is the direction that we’re moving toward.

Kerim: Right, kind of bringing that standardization to the industry where that point you made earlier, Armando, about like not making it a race to the bottom, making it a quality product, but built you know not, maybe with a little bit more extra cost, but will pay off in the maintenance years for the decades to come.

Jimena: Yeah, and I think like, oh –

New tech trends in monitoring and standardizing various solar development processes

Kerim: I remember in one of our previous conversations you talked a little bit about how you guys are really leveraging more and more of drones and robots in terms of EPC quality and the deliverables. 

Can you talk a little bit about the trends you’re seeing in terms of new technologies to kind of monitor, and also assess the quality of the jobs done at the construction sites, and how that is evolving?

Jimena: Yeah, so there are some great players out there, Raptor Maps, Terabase. There’s a lot of good technologies, and we’re starting to use it, one in terms of construction monitoring progress. And that is, sending out a drone and telling us how many piles have been installed, and how that reflects it in the percentage of the site built. And then on the O&M side, we’re more used to IR scans, which gives us the thermal signature that can identify whether a module is online, offline, or if it has a cell level defect.

With that said, where we’re going is we’re actually working on a pilot right now with autonomous drones. So what we’re doing is we’re putting drones on the site that are able to do repeatable missions as well as be called on demand.

So this means, we can do a monthly infrared survey that will tell us all the strings of line, and that can be constantly communicated to the O&M, but we can also deploy the drone in the event of a hail, in the event of, or even just to check, whether the site is stowed, right, before a big storm. I always joke about looking out the window, and somebody’s asking you, is the site stalled? And you look out the window and you say, Yep. But these sites are beyond our line of sight. So they’re just getting bigger and bigger. And you know, humans, it’s really hard to have that coverage. So we need to use drones. We need to use technology. And it’s also a way to quickly digitalize and have a digital record of the site that feeds into the digital twin, like, we said.

So yeah, we’re actively trying to identify what the best technology is. We’re working with the leading companies in the US to provide us these solutions. And I think a big part of this is also that our industry has grown so fast that everybody’s trying to do something new. And when everybody’s trying to do something new, it’s really hard to find the experts in their field.

Kerim: Seems like nowadays, like a lot of these data activities are driven by compliance and reporting requirements. How important is it to define those in the right way? And I think we also talked about Orange Button. Like what are some levers that industry leaders can play with or bring forth to create the standardization that you would like to see in the industry? How do you think about that?

Jimena: Well, I think a big one is GADS reporting. This is a requirement that actually started earlier this year. It was mandatory for any power plant north of 100 megawatts to report in a certain way, and it’s a very specific way that dictates what the plant ID should be, how to input the megawatts. That the data needs to be in 5 min, that we need to define certain issues a certain way. So I think some of these regulatory requirements are going to drive how we need to maintain our data to be able to fulfill this, but it’s really hard to get everybody to standardize across. Like I said, everybody’s using their own systems.

One EPC might use Procore. The other one might have their own internal software. So it’s really hard to maintain that currently. And like you mentioned, initiatives like Orange Button. They’re really trying to work to see what is the best way to have a single solar standardized data set across every single owner, asset, company.

How can we attract Data Scientists and Computer Engineers from Bigtech to Solar & Clean Energy?

Kerim: Yeah. Data problems, well, plenty of them, it seems like in your field. And we talked about how a lot of data-oriented young professionals are going to the Bigtech sector, but more and more the Bigtech is becoming energy-dependent. So perhaps, we can create those challenges in our industry to have them tackle the data challenges that our industry is facing. 

So on that, Armando, do you have any thoughts to attract those types of young guns and professionals and to — 

Armando: I mean, I feel like it’s one of the areas where our industry isn’t necessarily calling to the data scientists and the computer engineers that we tend to see going to Bigtech, right? And part of it is because energy isn’t always the most glamorous or —

Kerim: Yet, yet.  

Armando: Yet. 

Jimena: It isn’t?

Kerim: Yet. But this reminds me, this conversation reminds me of something Bill Gates used to say when I was first starting my career, 30 years ago in the tech sector. He was, back in the mid-90s, complaining that the brightest minds were going to Wall Street instead of going to tech. That changed about 15 years later. About 15 years ago, they started going into Silicon Valley instead of Manhattan and big Wall Street guys. A lot of bright guys started coming to Silicon Valley ecosystem. But I think the next phase of that, hopefully, could be our industry.

Armando: Yeah. One of the big things, to me is always, as an engineer, you know, one of the things that I value is seeing the product of what we’re putting into the ground and the longevity to it. A lot of tech and I don’t mean this to be in any way disparaging, but a lot of it tends to go up and go away fairly quickly, whether it’s apps or some startups, right? With energy, it tends to be something that has longevity, as well as, we’re always going to need it. As time progresses, the need for energy continues to grow. It’s not going away. It’s one of the pillars of our life, our industry, our future, and it really feeds to no matter what you go into it’s there. We need it, right? The lights are on in this room. The lights are on powering this recording. So it really speaks to that. 

So when it comes to a call to attract that talent there’s that opportunity of all of the unknowns that haven’t been really discovered yet, or really explored, because one of the things that really enlightens me about the ideas and innovations that Jimena, as an example, has brought forward, is how we can leverage these new technologies and providers, which is still even very early on, to make value of things, for our industry. 

The solar industry has been building projects the same for the past 15-20 plus years. A module gets connected, goes to an inverter, reports data how it’s generating. And that’s pretty much it, right? You’ve seen progress in inverters, modules, et cetera, tracking systems. But at the end of the day, all of that stuff stay the same. We’re still going through the same maintenance.

Jimena: Except the sizes. The sizes of the sites have grown substantially. I’ll say that because I have some mentors like the biggest site they had built a couple of years ago was 10, 20 megawatts. And now we’re hitting like 800 and above. So I’ll just say that, the size has definitely changed, and the like the speed of growth –

Armando: Yeah.

Jimena: – of our industry.

Armando: And that’s a good point, right? You’re still seeing that massive amount of land. But it’s obviously a lot more dense when it comes to the energy per acre. And I think that’s where we’ve seen a big part of that transition. But to the point of where what value individuals can bring in this new talent coming out of school can bring is looking at how they can make our industry more efficient. Right? 

We’re looking at putting in new solar parks, all across the US and globally. But we’re very quickly going to see that we’re going to repower a lot of these older parks to make them even better and bigger because the need for energy continues to grow, right? We’re “running out of land or really not but we can make the land that we’ve already used even more efficient.” Finding ways to make that cost effective is really in the data. 

And all of that starts from being smart about how we maintain it. Looking at how, what attributes we have with certain equipment, and how long it lasts, to not just speak back to any specific manufacturer and say, “This is a great product.” Or “Hey, I’d really like to see some improvements as we move forward into the next iteration and next version.” So it really helps all across the board our industry move forward. 

Kerim: What a great call out to data scientists who want to enter our industry. Thank you for that, for those last words, Armando. 

Challenges that lie ahead for data & computer scientists in the field of solar energy development

Any other thoughts that you want to add guys, as we come to the end of this conversation?

Jimena: I’ll just add that I think the biggest problem right now is that data comes in chunks, in different stages of the asset. But it’s still one asset. Right? And ultimately, where I think we should get to is, we should have a single record from when the site was first designed in the development process, and then that same record should be transferred over, and hold all the EPC data. 

And that means, what failed, what quality was flagged? You know, all those details that kind of get lost within all the PDFs and all the Excels, and all the punch lists, and all the work that’s been done. That’s not really captured. 

And then it is also where scopes start and end. You have a different scope from who designs the plan, and then, they give you all the information and then it’s on to the next, and then the next grabs all the documents, does their thing, finishes, and then it’s up to the next. But it’s still a single asset. 

So where I would like to see us grow is it just should be one thing. We should be able to tell what the development engineer was thinking. We should hold all the information of who built the plant. We should have all the operational data which is, we’re getting pretty savvy around it and how do we historize it?

And then we get we get to a place where I know Armando is really passionate about, and it’s the levelized cost of energy. And at the end of the day is, did we make the right decision, initially? And did that pay off in the operational side of it? Or did we make the wrong decision, initially? And you know, our projects are not as profitable?

So I think data should be, just encompass the whole life of the asset, and we should use it as best as we can, so that we can understand how to build better plants, how to make better decisions, how to not just buy the cheapest equipment, not just the initial, cheapest equipment at like face value, but to really buy the optimal equipment that will sustain the asset in the best possible way.

So I just, I really hope we get there because we are a tech. And speaking to everybody who wants to join, we are a tech. We’re using AI, we’re using SCADA systems. There’s controls. There’s data like, wherever you look at it, I’m working remote, and I can access a plant in Texas. That’s fun. And it is fun to be able to watch it live and to drive insights and to have information even being so far away. 

So we are a tech. And I think where we’re going is where it gets the most fun. There’s a lot of, like Armando said, energy is the pinnacle of our society, and our energy consumption is only growing faster that we can all catch up. And we need the AI to support this. We need the computer scientists to help us, you know, gather all, build the codes so that we can be program-agnostic. 

What I’m trying to say is, we are a tech, and we’re not just a commodity. We are like this industry is a necessity. So I think it’s pretty cool to be able to have all those insights and see, everywhere I look in the solar industry, people are driving innovation. People are driving insights.

I think everybody’s working with data to a fault because we can’t all agree how to unify that data. But we’re all data-driven. And I think I’m just really excited to be a part of this group and be able to be part of this transition where everything’s growing so fast. 

We need energy. And we need data to help us, data and technology, to kind of help us get there. Because there also is not going to be enough people to build our sites. And there’s not going to be enough people to be monitoring such large assets. So we need data, we need technology. And really, we need people who care about what they’re doing and who want to put passion and effort into having good reliable power plants to power our grid.

Kerim: Well, thanks. Thanks for that summary, Jimena and Armando. What a great in-depth conversation about the challenges of data in this very rapidly growing segment of the energy world. 

At this point solar is well into the mid single digit percentages of our energy infrastructure worldwide. And it is growing as we talked about very rapidly, probably 20% to 30% compounded annual growth rate worldwide. And with that comes all this data challenges for the largest and also small and medium-sized developers. 

This conversation has been an eye-opener for me in terms of the vastness of optimization and improvements that we can bring to the industry. And I hope that will bring more data scientists and tech people to help with these issues we’re facing as an industry.

Thanks a lot, guys. Thank you, Armando, Jimena, for this conversation. Yeah, with that, I will end this solar conversation, with more to continue. 

Jimena: Thank you.

Armando: Thank you so much.