Cet article est également disponible en français : Soleil, es-tu là ? (1/2)
The main advantages of producing electricity from solar energy is to benefit from a free and renewable resource; the operating price is lower than for coal or gas power plants for example and almost immune to changes in market prices. However, solar “fuel” has the disadvantage of being variable in time and space, and its quality isn’t as well characterized as that of conventional fuels. Yet it is essential to estimate an investment’s electricity production in order to assess its profitability.
In the case of photovoltaic installations, production depends almost proportionally on the energy received from the sun, hence the importance for project developers to have tools that evaluate the solar resource. Since the timeline for a project can span several decades, it is not essential to predict the power produced each moment, but rather to have an idea of the magnitude of the annual solar energy average that the production system will receive.
Evaluation of solar resource
Solar power that reaches the surface of the Earth’s atmosphere varies depending on the location (latitude, longitude),time of day and period of the year (hour, season) but can be estimated accurately at every moment; on average it is 1367 W/m2. However, the power reaching the ground is much more uncertain because the rays are reduced differently according to the variable constitution of the atmosphere’s layers. Small suspended particles called aerosols can diffuse a ray in many directions and clouds only allow part of the rays to go through them and they absorb or reflect the rest of them.
There are two methods to estimate the power received at the ground level :
1. Direct measurement on the ground
Pyranometers are thermopiles (converting a temperature difference into electricity) that are used to measure the power received by the sun. This is the best method for estimating the resource in one location because the results are more accurate, however, the geographic distribution of weather stations is uneven depending on the territory and the depth of data records varies from one site to another. Note that it is also possible to use photovoltaic cells as “reference” for the ground measurements, the incidental solar power being deduced from the electricity produced by the cell. However, the absorption spectrum of photovoltaic materials is narrower than the spectrum captured by pyranometers: the measured quantities are not strictly identical.
2. Estimation from satellite images
The principle is to apply an algorithm to satellite images to assess the incidental power on the ground. Several approaches are possible. For example, the HelioClim database developed at Mines ParisTech combines the theoretical calculation of the incidental energy in case of clear skies with the index of cloud cover. The cloud index is determined through the use images from the satellite Meteosat which covers Europe, Africa and parts of Brazil. In the United States, an equivalent database but one that is based on algorithms developed by Dr. Perez at the University of Albany is accessible through the SolarAnywhere platform.
This method has the advantage of producing consistent data over time and space, and maps of the average energy received over a certain period can be established.
The resource estimated by the two methods described is the power of the rays that reach the Earth on an horizontal plane. However, apart from the equator, the solar panels are slightly angled to maximize the amount of energy they receive throughout the year and they are oriented as much as possible to the south (or north in the southern hemisphere). Empirical models enable us to evaluate the resource in an inclined plane from the data known on the horizontal plane, but this increases the estimation error. An alternative is to place inclined pyranometers at production sites, but because of the cost this method is reserved for large power plants.
Moreover, thermodynamic solar power and concentrated photovoltaic technologies using concentration can only exploit sunlight coming perpendicularly to the surface of collection. This resource is called DNI (Direct Normal Irradiance). Some databases provide direct calculation of DNI from satellite images, like the German product Solemi.
Pyrheliometer is the instrument that enables measurement of the DNI. It is simply a pyranometer equipped with a collimator and placed on a tracker that evolves throughout the day so that the whole system constantly faces the sun.
Finally, the integration of data on power over time enable the evaluation of energy received over a certain period. At least ten years of data are needed to assess the annual average incidental energy in one location (by making the underlying assumption that future climate will be similar to past climate). In central Europe, the solar resource is about 1,100 kWh/m2 per year and it reaches 2,300 kWh/m2.year in the sunniest spots in the world.
Tools to access information
The US National Solar Radiation Data Base, which combines ground measurements and satellite observations, is the most reliable free of charge and recognized source for evaluating the solar resource. TMY3 (Typical Meteorological Year) files for more than 1,000 locations in the country are extracted from this source. Each file shows an hourly distribution of a plausible set of weather criteria -including Energy Solar- for a year. Another database, developed by NASA, provides the entire planet with twelve values of average daily energy (one per month) that are accessible through an Internet platform.
Finally let us underline the existence of interesting research projects on software tools aggregating data from multiple databases. At the National Renewable Energy Laboratory (NREL) for example, Geographic Information Systems help identify potential business locations for solar thermodynamic concentration (CSP) thanks to cross checking of maps of solar irradiance and maps of ground declension. In fact, the slope of the field should be less than 3% for the construction of a CSP plant. Other softwares from NREL offer the opportunity to evaluate solar output in one location, or even identify the production systems installed.
The Case of California
California is one of the sunniest states in the United States and is betting on green technologies, by conviction and to revive its economy. Therefore, it is logical that this highly technological region uses high-tech computer tools to generate information on solar energy. Some examples:
- The local government “Go Solar California” campaign aims at installing 3 GW of PV capacity by 2017. 893 MW are already in place according to the website updated in real time.
- Spurred by the “Department of the Environment”, an interactive map of PV installations was born in San Francisco. As shown in the figure below, there was at the end of May 2,043 solar roofs or 20% of the ambitious goal set for 2012 (10,000 solar roofs).
- Some electricity providers like Southern California Edison and San DiegoGas & Electric will publish maps indicating areas where grid connection costs are the lowest.
Consequences of a double-digit growth of the market for at least the past 10 years, photovoltaic power installed worldwide almost reached 40 GW at the end of 2010. The resulting production, 50 TWh, represents slightly more than the consumption of electricity in Portugal or Israel in 2008. In this context, it suddenly became important for the energy industry to know the distribution of solar energy on the planet and this is the reason why tools that allow access to historical databases have been created.
However, the estimate of the average energy received annually soon will not suffice, because the insertion of intermittent electricity could become a real headache for grid operators. In California, for example, where the objective is to achieve 33% renewable production by 2020, the California Independent System Operator (CAISO) is already concerned about solutions that must be invented and implemented in the coming years.
One solution to better control the balance between supply and demand for electricity at any moment is to use forecast models of intermittent production to establish production plans of dispatchable energy. But is predicting the weather an easy exercise? The prediction of solar energy, in relation to meteorology, is a subject that we will discuss in the second part of this article (to be continued).