By Gareth Brown, CEO, Clir Renewables
To date, more than 500 GW of solar PV assets have come online across the globe. This has been facilitated by the rapidly falling cost of generating energy via solar power and, as such, global solar capacity is forecast to grow to 10 TW in the next decade as more countries commit to decarbonize their energy supply and cut production costs.
To make the most of current capacity and forecast growth, it will be crucial for solar asset owners to ensure that their assets are consistently performing throughout each project’s predicted lifetime. However, it has become increasingly clear that the true potential of the world’s solar assets — and with it, the progression of the solar industry — is being held back.
The assumption that solar PV is a simple way of generating energy due to its lack of moving parts has resulted in a lack of investment in truly effective asset management, including both everyday operations and maintenance (O&M) and innovation. Unfortunately for asset owners, with this assumption of simplicity comes complacency, especially when it comes to collecting data from and optimizing these assets.
Earth, wind and solar
Solar O&M is often perceived to be as simple as identifying physically damaged panels and cleaning or replacing them. However, this reactive approach misses faults such as tracker errors and poor grid control — instead mistaking these problems for a dip in irradiance that are unavoidable.
Owing to the relatively static nature of solar panels compared to other methods of generating power, many owners and operators assume that there will be less wear-and-tear of parts and therefore less need to monitor for underperformance. However, solar panels are exposed to degradation from the elements and will suffer damage due to the ongoing influence of factors such as sunlight, different types of precipitation and the growth of nearby vegetation.
Unfortunately, much of this environmental exposure is not accounted for by the current industry standard for asset performance monitoring systems. Without a thorough understanding of whether underperformance is a result of environmental issues such as those outlined above, or even poor siting, asset owners mistakenly attribute low output to low irradiation or cloud cover, rather than a sustained and often-fixable error.
In order to accurately assess whether solar assets are underperforming and why, SCADA data must be set in its geospatial and operational context and fed through a common model that considers all available contextual and environmental data.
To properly understand the true causes of underperformance across a solar array, it is crucial to integrate current and historical meteorological data, surrounding forestry and vegetation, operational and grid activities, and SCADA data from an individual asset, then benchmark this data against the surrounding panels. These multiple data streams, when set in context of the project as a whole, will allow an asset owner to immediately pinpoint whether underperformance can be attributed to low irradiation or a technical error that can be quickly addressed.
By looking beyond single asset SCADA data and drawing from as many contextual data streams as possible, owners and operators can rule out assumed explanations for underperformance and zero in on the true cause.
Wading through the data
However, many asset owners and operators have avoided taking the necessary in-depth, data-based approach to optimizing their solar projects due to concerns about cost.
Under the more traditional, post-processing approach to data analysis, a vast number of working hours (and therefore wages) must be dedicated to wading through and analysing the volume of data required to effectively target O&M and improve performance — particularly when working with large-scale solar farms and international portfolios of assets. This increase in cost has actively disincentivised asset owners from using data to inform O&M and resulted in complacency around data collection.
This, in turn, has led to a negative feedback loop. Low-quality data collection results in technical teams having to spend more time relabelling and managing data, which prevents operations teams from relying on this data for quick answers and results in further complacency around data collection. We have now reached a point where low-quality data collection results in technical teams spending hours cleaning their data each day where they could be using their domain expertise to drive asset optimization.
This data quagmire becomes even more impassable for owners with assets spanning a number of different original equipment manufacturers (OEMs). As data collection methodology is different for each OEM, it is almost impossible to compare and optimize the whole fleet quickly.
In response to this challenge, asset owners have begun to adopt approaches to asset data powered by artificial intelligence. For example, Clir’s optimization platform integrates all available asset and contextual data sets across a solar project and at portfolio level, running this data through a common model. This effectively translates OEM data into a comparable form, allowing asset owners to immediately recognize whether a given asset is underperforming compared to its peers, regardless of the manufacturer.
A true understanding of the causes of underperformance — and therefore the potential for increases in performance — can open up a number of opportunities for solar asset owners. In addition to increasing annual energy production (AEP) from fixing current errors, owners who use AI-powered analytic technology can more effectively predict future AEP and prevent future underperformance. By moving away from outdated assumptions and using contextualized asset data as a foundation for modelling and managing risk, project owners will be in a better position when it comes to asset valuation if they wish to sell their projects as well as when negotiating with insurance firms.
In recognizing that solar power is not simple and investing in in-depth data collection and analysis, solar asset owners and operators can see significant gains in both the short- and long-term.
Gareth Brown is CEO and co-founder of Clir Renewables, a renewable energy AI software company. He is an entrepreneur, a chartered engineer with the IMechE, and has degrees in mathematics and mechanical engineering. Gareth has over a decade of experience in the industry which spans the life-cycle of renewable energy projects from identification, development, construction, to financing, and operation.