The economics of solar energy generation has steadily been improving thanks to continuous research and development, but its sensitivity the variability of available sunlight has always relegated it to a supplementary role. A new algorithm that can improve the size, positioning, and operation of solar generators may change this.

University of California in San Diego researchers Matthew Lave and Jan Kleissl have developed a way to predict fluctuations in the power output of solar generators when the clouds start rolling in. The potential benefits of being to predict how solar power generation fluctuates in advance of obstructions to sunlight can be significant. Utility companies who wanted to place bigger bets on solar can now refine their standby power generation requirements downwards reducing capital as well as operating costs.

The unpredictability of sustained solar power generation has prevented utility companies from using more solar generated energy. In Southern California for instance, utility companies that want to use solar power for more than 15 percent of their peak demand mix, are required to conduct extensive studies to ensure sustainability.

The algorithm works off a solar variability law discovered by Lave and Kleissl-the result of a total $2.5 million study grant from the Solar Energy Technologies Program of the US Department of Energy and the private sector which provided $500,000 of the grant which runs until next year.

And it is not only US government and private sector funding that is pushing for more efficiency from photovoltaic (PV) generating systems. In California a new law just signed by Governor Jerry Brown mandates an increase in the PV-sourced component of electricity supplied by the power grids to 33 per cent by 2020. The US federal government too is pushing for more efficient smart power grids that will allow for the increased uptake from renewable sources such as wind and solar while reducing use and dependency on systems power by non-renewable fuel.

Lave and Kleissl's discovery provides a way to improve planning and accuracy for how much consistent power PV arrays can deliver, as well as serve as the basis for creating accurate models on which planned expansion and increased PV use can be based. This reduces the amount of stand-by 'topping-up' capacity from fossil-fuel generators, it makes the deliverable capacity of PV arrays in various locations predictable so that needed supplementation can be sourced from renewable energy sources first before falling over to traditional sources.

The goal of the study is to be able to create a one hour window of predictability through modelling software that integrates a network of climate and cloud monitoring systems with a network of energy-generating PV arrays. The result is development of a system that monitors and predicts electricity demand and optimizes power sourcing from the cheapest available sources-being able to use more more renewable energy generators than currently possible.

The system and its related procedures is quite simple, according to Kleissl. As simple as going to Google Maps, looking for the unmistakable dark rectangular shapes on rooftops which are solar arrays, and then connecting them with lines to form a polygon. The shape they produce becomes the basis of a variability forecast that the algorithm generates, he said.

Of course, the researchers had the benefit of working with the most monitored power grid in the continental United States with 5,900 PV panels generating some 1.2 megawatts of power. By recording the amount of solar radiation each panel received for a year and how much their output varied when sunlight hitting them would be interrupted by as short as one second.

This resulted in the discovery of a solar variability law that established the relationship between the extent of power variability to the distance between PV arrays and the length of time sunlight to them is interrupted, Lave said.

Lave and Kleissl's work also made it clear that more efficient, PV systems can be achieved by building smaller arrays over a widely distributed area. Power variability is reduced in this case because clouds that affect one installation would not affect others.

At present, solar power variability is not a big issue because of the still low penetration of solar power in energy grids. But this will change in time because PV cells are not only getting cheaper, they are also becoming much more efficient. The economics will drive demand for their expanding use especially since their capital cost can be amortized over a significant time frame. As this happens, the need to study, monitor and predict variability will become increasingly important.