By Katie Applebaum, Business Development Manager, Clir Renewables
The solar power sector has grown significantly over the past decade. This growth is in part due to falling LCOE driving the development of new projects across Europe and North America. There are a number of reasons for this fall in LCOE, from falling silicon costs to growing capacity and the reduction in the use of expensive materials.
However, minimizing the development cost of a solar project does not guarantee its financial success. In order to maximize returns, solar project owners need to effectively manage solar performance via smart operations throughout the full asset lifetime — not just cut costs at the start.
Optimizing operations – lessons from other renewables
Once online, there are a number of ways that project owners can improve asset performance, optimize maintenance schedules, and ultimately ensure that projects are continuing to generate as much energy and revenue as possible. To discover these potential gains, owners must invest in understanding the hidden performance issues of their PV modules, inverters, and other components within and across projects at a portfolio level.
Other power sectors have been proactive in identifying how lifetime costs can be reduced and revenue increased via project data. In particular, the wind industry has recognized the value of analyzing turbine performance closely to ensure that the assets are behaving as expected in line with environmental conditions and resource. By analyzing turbine performance patterns for optimal energy generation, wind project owners can more effectively monitor asset health and manage financial risk.
In-depth data monitoring and analysis have therefore become central to identifying and labeling reduced performance of wind assets. Once issues — misapplied derating, a lack of wind resource or mechanical failure, for example — are recognized through the asset data, wind project owners and operators can improve performance via controls corrections and upgrades, plan downtime and repairs, and accurately forecast annual energy production.
While solar PV plants have a fraction of the moving parts that wind farms do, they still produce a wealth of data that can inform maintenance plans and signal routes to increasing annual energy production. However, uptake of data monitoring with in-depth analysis has been slower than in wind.
Too much data to handle
In comparison to other power generating assets, solar PV modules and accompanying components can be cheap to manufacture. As such, the financial barrier to entry for new manufacturers is low. These low prices have resulted in a wide range of PV module and component manufacturers competing in the global market. The number of component manufacturers is a stark contrast to the wind industry, where the scale of the technology means that R&D and manufacturing can run into the hundreds of millions of dollars — resulting in the consolidation of manufacturers to five main players in the European and US markets.
The proliferation of manufacturers across solar technology has introduced a further complexity to handling PV data: the wide variety of ways in which different equipment manufacturers format and report performance data across otherwise comparable assets at a typical PV plant.
Varying data formats at PV projects makes an immediate like-for-like comparison — let alone immediate notification of a problem with a single component — challenging. Herein lies the problem: to discover opportunities to improve solar asset performance across varied portfolios, asset owners need to be able to quickly and clearly compare projects and benchmark their performance, regardless of the component manufacturer or series.
The variation across solar OEM data is a particular problem for companies with multiple projects consisting of equipment from different manufacturers in their portfolio. At present, to compare and benchmark the performance of the components of each project to identify potential gains, owners must first invest substantial time into translating many different datasets into a standardized format. As such, many owners are discouraged from intra-portfolio comparison, since the cost of continually translating this data often outweighs the perceived value of asset optimization.
Translating data into a common model
To tackle this issue, Clir is working with renewable energy investors, operators and owners to create a standardized, manufacturer-agnostic data model for PV plant data.
Manufacturer-specific datasets are ingested into this standardized model, to which Clir’s machine-learning driven suite of analytical tools are applied to provide insight into asset performance. Contextual data such as local irradiance and other relevant climatic conditions are also incorporated into the common model, enabling asset performance to be analyzed in the context of resource and environmental interference.
For example, the model can carry out soiling analysis based on component data, weather reports, and historical patterns in these datasets, which can then inform cleaning schedules to prevent dirt from building up to the ratio that can block out the panel’s input.
The highly competitive solar PV equipment manufacturing industry has been key in driving down the upfront costs of PV plants. However, the multitude of manufacturers and innovative products need not be an impediment to realizing performance gains via project data analytics once projects are operational.
Machine learning can cut the time necessary to translate and analyze data to hours rather than weeks, providing analysts with a full understanding of project performance in relation to the whole portfolio when it is needed, rather than to reflect data from weeks or even months ago.
With this information to hand, project owners can realize maximum benefit from the low start-up cost that the proliferation in solar manufacturers has brought, without losing the like-for-like benchmarking and scalable analytics that are crucial to informing financial decisions for existing projects.
Madan LAL SACHDEVA says
It is welcome move for analytic of data monitored from solar panels including inverters, combiner, reactive unit, etc. but what will be its cost effect on panels due to introduction of AI, ML .etc. The solar market has gone very competitive and any cost addition may not be accepted by the buyers.