Skip to main content

Photovoltaic and Storage System Cost Benchmarking (Text Version)

This is the text version for a video—Photovoltaic (PV) and Storage System Cost Benchmarking—about how to use a bottom-up analysis methodology to model costs for PV systems.

It’s Part 3 of NREL’s Solar Techno-Economic Analysis (TEA) Tutorials video series.

[Audio begins]

Hey, everyone. Thanks for joining us today.  In this part of the presentation, I’ll be talking about our USA solar and storage cost benchmarking, and the concepts and approach behind our benchmark numbers on a very high level.

Agenda

With this presentation, you will get to know how our PV and storage cost benchmarking works, and I’ll also discuss our preliminary results of QN2020 solar and storage benchmark numbers. And I want to point out that benchmark estimates for this year doesn’t include any cost impact due to COVID-19. However, we will see how COVID-19 has affected the industry as a whole at the very end.

System Cost Benchmark

Just a little bit of background behind the work that we do at NREL. NREL has been benchmarking solar and storage costs since 2009. The cost models are a bottom-up approach where we aggregate all the different system and project-level costs based on industry standards for a typical project size in residential, commercial, and utility sectors.

Our total system cost does include key soft costs like EPC and developer overhead and profit margin. The main goal of our work is to help the industry and policymakers understand the cost reduction opportunities and future focus. Our total cost estimates are all based on ballpark with the industry estimates, though it might vary state by state. Sometimes they do not match with other industry estimates because of the way the models are structured, and the costs counted in each category could be very different. So, it is always advised before comparing any two estimates, it is imperative that we understand the models and also the way the costs are categorized in different buckets.

Key Cost Components

So, these are our key cost components, or categories broken down from our total system cost. So, the first two cost components you see are commodity components, and then the highlighted cost components are where we have done a detailed modeling by analyzing all the different sub-component activities and labor effort required, things like site preparation, racking, and writing activities, which are typical in a solar or a storage project.

Also, we have to take into account how the costs and quantities of these sub-components or activities change as we introduce economies of scale in our modeling, and also how these activities could get affected for different system designs. For example, DC couple energy storage could have different sub-activities compared to an AC couple energy storage. And in the bottom of the presentation, below the highlighted area, are the soft costs, and these soft cost estimates are usually gathered from industry interviews, or from third-party research firms, which have published their quarterly report.

High-Level Framework

This is our high-level framework for our system cost modeling. Basically, in order to calculate all the key cost components, we need to list out all the major steps involved in a PV installation project, like the highlighted, blue-colored borders here. Also, we need to figure out different inputs required, and then set up the calculation depending on your benchmarking needs.

For instance, in this case, in order to estimate cost of PV installation project, the major items we would like to figure out are how many number of modules and inverters required, how much space is available for installation, how much material we would need for racking structure, and how much labor it takes for different construction and electrical activities. So, it is important to collect all the inputs in order to estimate all the major functions in the aforementioned activities. For example, we need location-based inputs like wind and snow. As we can see in the middle orange box that shows, you know, different inputs by location that are required. So, these inputs determine number of vertical columns and spacing required to support the tubing structure with the expected wind and snow load for a specific location.

After you have all the activities and inputs figured out like this, next step is to set up the calculation for each step and assign it to a cost category. At the bottom of this slide, we have an example, and here, as you can see, we have three different categories listed out as an example with different activities under each category. So, these activities would change depending on the system configuration, or, you know, depending on the project’s specific need. So, these could get as detailed as circuit, depending on the project’s needs, or the benchmarking requirements. And, after that, you have to figure out the quantity and the material and equipment and the labor-related costs and units, and we use construction costs data from RS Means, which is a popular source to have estimated construction cost related data. So, we use that for a material, equipment, and labor effort estimation for a particular activity.

And, you know, we use Bureau of Labor Statistics and use their labor wages for all these activities. And for commodity prices and soft cost, we usually depend on, you know, the quarterly report from third-party research firms like Mackenzie or Bloomberg, and also, sometimes we get the data from industry interviews.

Python Model Structure

Initially, we developed our models in Excel, but during the process of transitioning completely to Python, due to issues like performing complex calculations, handling large data sets, readability of the cost model, and also, I know, issues due to Excel crashing often. So, with the Python model, we have automated use of certain data sets where the data has been automatically scraped from corresponding websites and stored locally in our repository for calculations based on the benchmarking year.

This is a high-level diagram that shows our Python model in a gist. But, you know, it gets easier when it comes to readability and that’s one of the main reasons we had moved from Excel to Python so that anyone who use the model could understand the functions and different components of the model easier than before.

Q1-2020 Utility PV Model Inputs and Assumptions

This is just the assumptions and its associated source of data. I’m not going to get into details, but followed few important assumptions and sources. One of the main sources for our assumptions comes from Lawrence Berkeley National Labs’s Tracking the Sun Report, our model efficiency DC/AC ratio inverter market share, and also, you know, the market share between small installers like third-party owned parties versus large integrators.


All this data comes from LBNL’s report for this preliminary analysis. However, the module invertor and commodity prices and other soft costs are based on our interviews we have conducted with the developers and other independent R&D firms.

Q1-2020 Utility PV Model Preliminary Results

In 2019, we have used multi-crystalline silicone PV and its related assumptions in PV cost estimates. So, in this slide, you could see three different scenarios, 2019 multi-silicone PV cost estimates and 2019 USD. And, likewise, you have both mono PERC estimates in 2019 and 2020. So, due to increased market share of mono PERC of late, we have used mono PERC modules for our cost estimates. So, in order to have apples to apples comparison this year, we also modeled a scenario with 2019 cost estimates using mono PERC PV crystalline modules.

As you can see, between 2019, multi and 2019 and mono PERC scenarios, costs are getting really competitive, suggesting a shift in market preference, even the third mono module price in 2019. However, when you compare the 2019 estimates with 2020, the price has reduced, and the main drivers of cost reduction are module efficiency. Of entire module efficiency, we need lower number of panels for a given project size, and, thus, it helps us reduce the size and the components of different structural and electrical activities, and also, you know, associated labor recommended to those activities. However, drivers of cost increments are relatively a little bit higher module price, and also it is worth it to note that the labor wage has increased over the years due to inflation. Though we have reduced labor costs in total, due to improved efficiency of the module, the labor wage has increased over the years because of inflation.

Q1-2020 Utility-Scale Storage Preliminary Results

So, moving on to next slide. So, these are our 2020 utility-scale, stand-alone storage numbers, which are yet to be finalized. All of the standalone batteries are assumed to be AC coupled, and we assume a 60-megawatt system of different durations for our utility-scale battery estimates. Our battery pack cost varies with duration this year, and in previous year estimates, we always had a same dollar per kwh for our battery pack costs. We do have PV plus storage models and estimates, but due to time constraint, I haven’t included our preliminary estimates of PV plus storage here. But, I can discuss some of the key differentiating factors in our PV plus DC-coupled and AC-coupled storage configurations. So, our first key modeling assumption with DC-coupled batteries is that they require less additional balance of system components.

For instance, we don’t need an extra bidirectional model on the AC side in a typical DC-coupled configuration and its associated balance of system components, like switchgear or transformer are also not required. So, thus, it requires a little upfront component cost. However, the electrical wiring and its related labor in a DC-coupled system could be more as DC batteries often have a large number of smaller battery packs with charge controllers. So, when these batteries have, you know, smaller, so you need more labor effort to collect all those battery packs individually. And, also, when the batteries are located in a completely different site, it drives up the soft costs like, you know, collection and inspection. And, also, it increases the electrical work since the projects are literally happening in two separate sites.

Although the DC systems have a better roundtrip efficiency compared to AC-coupled systems, when it comes to retrofitting projects, or, you know, a project that requires simplified installation setup, AC-coupled systems always tend to be, cost a little bit cheaper than the DC-coupled systems, even though there is a necessity for an additional bi-directional inverter.

Q1-2020 PV Cost Benchmark Preliminary Results

So, this slide has summary of our residential, commercial, and utility PV costs for a standard project size in all those sectors. In 2020, we are seeing a five percent reduction in our preliminary estimates compared to our 2019 numbers. Major cost drivers across all these sectors is due to increase efficiency, because of the mono PERC modules that we have used in our cost modeling assumptions.

As of now, all of our estimates are in ballpark with the industry average, and as I previously mentioned, one of the main differential cost components is the way the soft costs are accounted, you know, in different buckets. So, it is always important to consider how the soft costs are modeled when you compare two estimates, as I said before.

Q1-2020 Storage Cost Benchmark Preliminary Results

And these are our current 2020 storage cost estimates for our batteries across all the three sectors. Interestingly, our estimates align with other reported values so far, and based on the interviews, I believe the difference in cost estimates when it comes to battery storage is mainly due to the way the model is structured or set up. In order to better align the estimates for lower duration batteries, we are actively improving our model, accounting for factors like depth of discharge, roundtrip efficiency, cycle life, and how they could impact the system cost upfront. And, also, we are trying to model how the balance of system costs vary with duration and the, you know, different factors I mentioned before.

COVID-19's Impact

Let’s look at how COVID-19 has impacted our industry as a whole. When it comes to utility-scale projects, there is not a huge impact, like the commercial and residential sectors are facing the impact. While 2020 installation estimates for utility-scale projects has reduced in number, it is still higher than the 2019 installation capacity. But, for residential and commercial sectors, the revised installation capacity estimate has fallen anywhere between 30 percent and 50 percent.

So, why have the 2020 estimates reduced? You know, the utility-scale projects are mostly in remote areas, and in most states, the construction activities were considered essential during the pandemic, so the projects resumed as usual, only with certain projects having to be postponed. So, even though they show a decline in value of installing capacity, some experts are saying utility-scale projects, you know, might happen as planned, as the year began, when the year began, and they might be able to meet the previously estimated capacity.

So, however, the major factor that concerns utility-scale projects is the timeline shifts, as some of the developers are reporting. But, one other factor is the increased overhead due to the increased warehousing cost because of the timeline shift of the projects they’re expecting.

However, residential and commercial projects are heavily impacted due to the fact that, you know, homeowners or business owners are either trying to conserve cash, are not willing to have a crew come, a crew from outside to the installations in their homes. So, this also leads to the assumption that the scenario could reduce the soft costs across the residential and commercial sectors with some of the activities moving online rather than happening in person. So, this might be something like having permitting and interconnection processes taking place online and targeting potential customers through digital channels.

So, this might also lead to increased labor due to the fact that, you know, businesses are trying to reduce their overhead costs, and owners are not willing to have too many outside crew members working on their home. So, this is still under evaluation, and these things are evolving as we get to know more about the pandemic. And, hopefully, with no second wave in all the countries, the effects of COVID-19, you know, reduces. So, that’s all I got for you guys today, and if you may have any questions, feel free to email us, and we are happy to answer any questions you guys may have. Thank you so much and have a good rest of your day.

[Audio ends]

To continue with Part 4 of the Solar TEA Tutorials video series, see Levelized Cost of Electricity and Internal Rate of Return for Photovoltaic Projects (Text Version).