Dr. Xavier Le Piver, CEO of the French start-up Steadysun, shares his expert opinion on the forecasting of solar electricity production and the new efficient technologies now available on the market. This article was first published on March 2nd 2015 on Actu-Environnement.com, the French B2B news portal of the energy sector.
We see a real global momentum that boosts solar energy. As its cost is becoming competitive, the growth of solar energy becomes exponential. Solar energy could become the first source of energy within a few years . Solar energy remains hampered by many technical challenges that need to be solved in order to facilitate its integration into power systems. Among the current solutions, one is easy to implement and it is also the cheapest: forecasting the production of electricity from solar sources. New combined approaches (physical + mathematical) demonstrate their effectiveness on site. Highly innovative, these new forecasting technologies are not recognised enough.
The solar market moved up a gear
Yes, the solar energy market has matured very fast! Apart from any environmental consideration, and from a purely energy point of view (annual production), it is often easier and more profitable to build a solar power station than a thermic station. This dynamic is leading mutations in the energy strategies of many countries and operators. In some regions of the globe, large-scale deployment of solar power is bringing down solar energy cost for the users. Implemented solutions are gaining daily in reliability, performance and profitability. This causes a domino effect. In many countries, grid parity is reached: renewable energy becomes more competitive than traditional energy. Public incentives then stop and let the market take over.
But technical hurdles remain
It is difficult to massively integrate solar energy into the power grid. By definition it is a variable energy source, which must be injected into stable electrical networks that should remain stable. Indeed, the electrical system is based on a set equilibrium. What is produced must be used simultaneously. A breakdown in this equilibrium (production/consumption) is a real nightmare for the network manager. For example, a cloudy weather provokes a screeching halt in production, which can be a source of network instability. Intermittent solar production is therefore a risk that the system operator must hedge.
To overcome this potential risk, operators have developed various strategies. They can use regular production capacity, in house or from the energy market. In a dynamic manner, operators also have the ability to limit solar production level or increase backup, with time constrains that can range from seconds to hours. Nevertheless, these compensatory actions cost money and it cannot be ignored.
How to integrate more solar energy in the network?
Four solutions come out:
- strengthening of network interconnection,
- modulation of consumption,
- energy storage,
The first 3 solutions require such an enormous investment that the unique objective of integrating more solar energy is not profitable enough. These solutions are profitable only when several functions are fulfilled and tackle several issues. Note that energy storage is expanding thanks to a plethora of new technology tests. However, it remains expensive. The last solution – forecasting the production of solar power plants – is easier and less expensive to implement. Forecasting solar production comes as a surgical tool, lightweight and precise. It requires few investment and now has a huge potential thanks to its accuracy. Combined with other solutions, forecasting production of solar power plants is a real lever of development for solar energy.
More powerful forecasting technology
What is new is the unprecedented level of accuracy due to the combination of several approaches. Physical models improve traditional mathematical models based on artificial intelligence and statistics, and a wide variety of types of information are used. This solar forecasting approach has been developed in France and now provides operational results. The implemented solution indeed uses several sources of information over different spatial and temporal scales.
Three levels of complementary information are combined. First, weather forecasts from rather large spatial mesh (50 to 10 km and even locally down to 1 km) have updating periods from 6 to 12 hours. In addition, satellite image forecasts are used from finer spatial mesh (between 3 and 1 km) and are updated every 15 minutes. Finally, we can combine hemispherical images captured on-site in real-time using video cameras specifically developed for this application. These video cameras can cover areas of several km² with high accuracy.
Depending on the circumstances and needs, forecasting accuracy can thus be increased over longer or shorter periods of time. As a consequence, uncertainty can also be greatly reduced.
Solar production forecast: a rapid return on investment
Managers of solar power plants equipped with such innovative systems are reducing ratio error from 40% to 50% compared to single forecasting systems. Thus, it also allows uncertainty cost reductions. Leading to considerable business opportunities:
- Within interconnected networks, where MWh sells between 40€ and 60€ on the energy market, the uncertainty linked to photovoltaic production costs between 2€ and 3€/MWh. On average, we can allow a saving of around 2,000€/installed MW per year.
- Profitability is even more significant on islands (and isolated sites) where thermal MWh costs between 200€ and 300€ against 50€ to 150€ for solar MWh. In these insular situations and more generally for hybrid systems, the more photovoltaic production is integrated, the lower the bill to the end customer is. In such situations, we observed uncertainty costs of 10€/MWh, and savings reaching up to 20,000€/MW installed per year!
Every day, most professionals in the solar sector who discover this new production forecast approach incorporate it into their implementation roadmap.
Steadysun is a start-up with 8 employees founded in 2013 by Xavier Le Pivert. It provides solar production forecasting services of unprecedented accuracy over horizons ranging from a few minutes to several days. Its solutions are used in France and around the world by electrical grid operators, solar plant operators and other players in the electricity market, over very diverse geographical areas.
 According to the International Energy Agency (IEA) in September 2014
 Rate widely used in Europe