2021 marked the third year in a row that solar represented the largest share of new generation capacity in the United States. We should be proud of this feat, but our work remains cut out for us; continued supply chain constraints, tariff investigations, and inflationary markets all pose uncertainty to development pipelines in 2022. The solar industry is no stranger to adversity. Previous dips in the “solarcoaster” have trained our industry to exercise muscles of creativity and collaboration. Those muscles will be critical this year to identify new solutions to industry challenges. This year’s Solar Risk Assessment is another testament to the willingness of industry’s leading experts on measurement and management of solar risk to collaborate and leverage data and science to move our industry forward. Designed intentionally for the non-technical solar financing community, this report has been and will continue to be refreshed every year to provide the latest insights on the evolution of solar risk.
92% of lost EBITDA is due to underproduction, dwarfing all other sources of risk Underproduction continues to be the greatest source of EBITDA volatility. Conceptually, the two biggest risks in any solar loan is a loss of revenue or an escalation in operating expenses. However, an analysis of our database (covering >30% of the US operating fleet) reveals that the average project has not experienced a material increase in total operating expenses compared with the budgeted amount at financing. Meanwhile, widespread underproduction is the largest cause of EBITDA shortfalls, with 92% of all lost EBITDA caused by poor availability, unrealistic production forecasts, and lower-than-expected irradiance.
Last year’s 2021 Solar Risk Assessment (SRA) report revealed that solar projects are experiencing P99 events once every 6 years instead of once every hundred years, 17x more often than expected. In an ongoing review of the data, we see these trends mostly continuing unabated. In addition to yearly underperformance, 1-in-7 projects are persistently underperforming their P99 over multiple years, highlighting that underproduction risk remains exponentially greater than expected.
New build solar capital expenditures have increased by 8% for a typical 100 MW PV system compared to 2021 PV system prices continue to remain high through the first half of 2022 as supply chain constraints and policy uncertainties create a high-risk environment for development. However, improvements in module technology and the adoption of large format modules will provide some relief via amortization of both component and soft costs. The pandemic heavily disrupted the global PV supply chain, leading to increased freight costs and longer lead times. Elevated commodity prices, such as steel, aluminum, and copper −which will be prolonged due to the Russia/Ukraine conflict − continue to squeeze already-tight developers’ margins. Labor shortages driven by the ‘Great resignation’ from 2021 will drag on 2022 and increase development costs. Other soft costs, such as profit margins, contingencies, insurance costs, and project management fees have also increased due to the risks associated with project development in this uncertain environment.
Probable tariff exemption will allow US solar to exceed 24GW this year, but supply constraints remain The US Department of Commerce said earlier this year that solar silicon cells made in Malaysia, Vietnam, Thailand and Cambodia may be subject to new antidumping and countervailing duties. More than 85% of module imports in the US last year and roughly half of the cells came from the targeted countries in the tariff proposal. Without knowing the level of potential tariffs and because they may be retroactive, major module makers halted shipments to the US. But on June 6, most of the US solar industry breathed a sigh of relief after the White House effectively granted Commerce the authority to hold off for two years on imposing new duties on solar cells and modules from Southeast Asia. The exemption would hold even if Commerce finds duties to be applicable in its preliminary results expected in the next few months. Citing a 1930 law and concerns over short-term grid reliability, the move appears to bring at least a temporary end to disruptive tariff uncertainty that delayed dozens of solar projects.
The Biden administration’s decision comes as a big relief for US solar developers. We expect developers to pile up modules in the next two years, following the example of Indian buyers who collectively stockpiled almost a year’s worth of build before a new tariff kicked in on April 1, 2022. That said, US module buyers would be wise to keep a close eye on the courts. Auxin or another party could sue the federal government over the 24-month exemption, arguing that it acted unlawfully. The administration appears to have applied the 1930 Tariff Act in a novel way.
In summary, losses attributable to terrain will be higher for sites with more complex topography and will be a key point of investigation for project developers who wish to evaluate the cost versus benefit of implementing advanced backtracking algorithms for their projects. DNV implemented the methodology for complex terrain evaluation in early 2021 as part of its baseline methodology.
Commercial and utility-scale PV systems are degrading at – 0.75%/yr after accounting for availability issues The long-term gradual performance loss of PV technologies is of great financial importance. Many studies have focused on this topic, yet, confusion often surrounds the subject because of module versus system and recoverable versus nonrecoverable losses. In the PV Fleet initiative, high frequency data from a large number of commercial and utility-scale systems is collected. To date, more than 1,700 sites, with ca. 19,000 individual inverter data streams provide a high-level perspective of the long-term performance of PV systems. The total monitored capacity has surpassed 7.2 gigawatts (GW) or roughly 6% of the entire commercial and utility market in the US. The open-source freely available software package RdTools is used to determine the performance loss rate (PLR) of the system in the database. The aggregated PLR distribution from the entire fleet is displayed in blue in Figure 1 (a) with a histogram of individual inverter-level PLRs. The median PLR for the fleet is found to be -0.75
%/year based on 4,915 inverters passing automated data quality checks This is lower than other recent publications (e.g. Bolinger 2020) that have established higher system PLR values around 1 %/yr. Because of our high frequency data we are better able to account for availability issues. It
is also possible that a difference in the portfolio or makeup of systems could include faster degrading systems. In addition, soiling deposition can contribute to annual performance loss even if cleaning events bring the system back up to full production. Identifying and isolating soiling performance loss trends remains an active area of work for the PV Fleet initiative.
A second histogram is shown in red displaying previously published literature values from our 2016
degradation summary paper. The vast majority of the data points in that previous study were based on
module-level degradation and did not present system loss. The example in Figure 1 (b) illustrates how
recoverable (fixable) losses such as inverter outages, stalled trackers, outages due to fuses, breaker or even curtailment, can explain a higher PLR than module-level degradation. Because of the high-fidelity of our data, we can now better account for availability issues and obtain a more accurate estimation of system losses. It is essential to differentiate between module and systems losses and use the appropriate
number for the most accurate financial models.
Wildfires caused up-to-3% annual soil–related performance losses at solar PV sites in California between 2018 and 2020 Wildfires have major detrimental effects on the performance of solar PV arrays in nearby areas in two ways. They make atmospheric conditions smokier or hazier; and they also deposit particulates on the solar PV panels, known as ‘soiling’. This reduces the sunlight reaching the panels.
Research from Power Factors shows that the wildfires in 2020 led to excess soiling-related performance losses of between 1% and 3% annually at affected sites. We learned this by analyzing data from 150 solar PV arrays that represent 1.3GW of AC capacity in California from 2018 to 2020. Operators can clean the panels to mitigate the effects of soiling, but they need accurate project performance data to know when this makes financial sense. Degraded Performance Operators can decide when it makes sense to clean soiled panels using the degraded performance methodology of Power Factors. This looks at the actual performance of solar PV projects compared to their theoretical performance, and shows operators when projects are underperforming. Further analysis can show operators the likely reasons that the projects are underperforming, by comparing their performance to signatures that show how four common types of degradation affect project performance. For example, underperformance caused by dirt, dust or ash will result in steady drops in project performance over time until panels are cleaned. It is a different profile to underperformance caused by snow, combiner outages and stalled trackers. This shows when soiling causes losses.
Wildfire Impacts Our research showed that soiling rates were significantly worse in 2020 during months when wildfires were at their worst (Figure 1). It shows a clear correlation between wildfires, soiling, and underperformance. The research showed that projects most affected by soiling were in counties in California most affected by wildfires, such as Kern County.
Days where wildfire smoke impacted solar doubled in 2020, 2021 compared to historic wildfire years of 2017, 2018 According to a new analysis by Clean Power Research, California’s solar potential was down 17% in September 2020—the peak of the worst wildfire season in recent history—relative to the long-term average for the month. Smoke clouds and aerosols from wildfires block sunlight and thus reduce PV output. Across California, resulting losses were 27 kWh/kWDC. Increased PV soiling from soot and ash also reduced yield (Clean Power Research estimates a median loss of 3% based on models utilizing concurrent particulate matter and precipitation data). Total impacts exceeding 30% were seen in regions surrounding large fires (Figure 1).
For stakeholders needing to forecast solar yield and asset value, observations from recent years provide new information on the risks to solar projects. First, some locations will be more impacted by smoke than others. Intuitively, proximity to wildfire fuel increases risk. But in addition, it’s been observed that smoke traps itself in valleys by blocking sunlight and air circulation, creating a self-reinforcing temperature inversion. The effect is apparent in the California Central Valley, the Columbia River Basin, the Po valley in Northern Italy and the Sichuan Basin in China, for example. Second, there is increased risk to solar-powered generation during wildfire season throughout Western North America. Interestingly, annual deviations for both 2020 and 2021 were moderated by unusually sunny spring weather. A shift in the seasonal yield profile is possible. NOAA’s seasonal outlook predicts that June – August 2022 will be hot and dry. The stage is set for another active fire season. While we will hope for the best, stakeholders must prepare for the possibility that wildfires will materially impact solar assets for the foreseeable future.