According to Capgemini’s annual World Energy Markets Observatory study, renewables were the fastest-growing worldwide energy source in 2018, with an increase of 14.5%. As renewable energies become increasingly important, traditional business models are under pressure. Increasingly, to support this shift, energy companies are moving to new models that include energy marketplaces, wholesale energy-trading services, microgrid-as-a-service, and local peer-to-peer energy exchanges.
Solar photovoltaic energy is part of this transformation, and represents almost 2% of worldwide energy production. Due to its intermittent nature, solar can sometimes produce excess energy, and where there is excess, there is trading. Excess power is transmitted to the grid, where it can be consumed by other users. This occurs in what is known as a peer-to-peer energy exchange. This opportunity presents a win-win situation for the buyer, who is able to purchase power at a cheaper rate, and the seller, who earns money by selling energy which otherwise might have been underutilized. Though we cannot yet capture sunlight in a jar, we are coming close.
This process, while beneficial, is not without risk. Because grid capacity is limited, the injection of excess energy increases the chances of grid failure. However, the emergence of new technologies, such as machine-learning models and predictive analytics, can help electricity utilities enhance grid safety without having to invest in increased capacity. Machine learning models can predict solar energy generation by considering various inputs such as weather parameters (i.e. radiation, temperature, wind velocity, etc.), solar-system characteristics, and operational activities.
It also helps organizations predict accurately to avoid overestimating and end up paying penalties associated with contract non-fulfillment, or underestimating and wasting energy that could have otherwise earned revenue.
Accuracy is key, but accurately predicting the amount of solar energy that will be produced is notoriously difficult given strong variances in solar radiation depending on location, season, landscape, and weather. It’s also important to understand that cleanliness and efficiency of solar modules and equipment have an impact on the amount of energy production.
For example, we know that dust on solar modules decreases performance by 10% and that the modules degrade by about 0.5% per year. Together, these variables make it possible to accurately predict how much solar energy will be generated. This not only eliminates commercial risk from energy trading, it also helps streamline grid operations. And though this will not help us capture sunlight in a jar, we can better harness the power of renewable energy moving forward.
Jagtap Kunal and Shreya Chopra are part of the application and cloud technology team at Capgemini.
Shreya is a management consultant working across multiple business units within Capgemini to solve the needs of our diverse set of clients through the use of the right technology.