You are here
Solar Power Simulation Engine
Sunlight: The Foundational Resource
Solar power is on track to becoming the predominant fuel for the clean energy economy. Unlike other forms of renewable energy, sunlight is generally ubiquitous, and production can readily scale from utility-grade down through industrial and commercial to residential and even portable. Yet some of the same features that make sunlight so promising a resource present some of the most significant challenges to developing large-scale generation facilities, managing distributed rooftop production, integrating with existing infrastructure, forecasting and operating profitably over the long-term.
- Average Incident Solar Radiation (“Insolation”) in kiloWatts per meter squared (“kWm2”).
- Size of the site in acres Number and average size of parcels that comprise site Proximity to infrastructure such as roads and transmission
- Capacity of nearest and best transmission
- Slope and grade of site
- Ground cover ratio
- Cost and benefit of trackers
- Day and night average high and low temperatures
- Cloud cover and other climate factors
- Frequency of rainfall and amount of dust and dirt
- Brownfield designation
- Local Rate REP, feed-in tariffs, renewable energy incentives and tax credits
- Best type of photovoltaic (“PV”) or concentrated solar power (“CSP”) to deploy
- Total power potential of site, as raw insolation and as derated production output estimate
- Revenue and profitability potential of site
Additionally, it’s important for developers to consider only the sites most likely to be productive. For this reason, Geostellar excludes sites from consideration based on such criteria as parcel size, zoning, environmental restrictions, hazards and buffers around infrastructure. As the project developer leases, permits, finances and constructs the production facility, the same core model that led to the selection of the site is used to support accurate production and operational simulations. Short-term weather forecasts are applied to estimate high and low production potential for a day, week or month, at any time of the year. The simulation quantifies risks over the next decades as the area gets built out with other renewable energy sources, incentives expire and tariffs change. Geostellar’s integrated solar power production model is a significant advance over previous methodologies.
Satellite sensor measurements
Of the many factors that must be considered when selecting a site for utility-scale solar power production, insolation is the most significant. In any particular region, the developer needs to locate the sites that will reliably provide the most power on the least real estate over the life of the facility. By combining datasets derived from satellite sensors and fixed-wing aircraft, Geostellar generates accurate ten square meter resolution insolation ratings over whole continents as a foundation for site selection and development-related decisions.
Geostellar’s insolation rating begins with the most advanced satellite measurement datasets available. These were developed by the Atmospheric Sciences Research Center (ASRC) at the State University of New York (SUNY)/Albany and the United States Department of Energy (DOE) National Renewable Energy Lab (NREL) using imagery collected by the Geostationary Operational Environmental Satellites (GOES) launched and maintained by the US National Oceanic and Atmospheric Adminstration (NOAA). These sun-synchronous satellites remain at a fixed point above the Earth’s surface over the equator, collecting visible-channel images of the entire hemispheric field of view every 30 minutes.
Annual PV Solar Radiation at 10KM resolution
The ASRC method converts the visible channel imagery to hourly estimates of solar resources on a 10 km grid, including direct normal insolation (DNI), used for assessing production of concentrated solar power (CSP) and global horizontal insolation (GHI), which, when combined with DNI, produces latitude tilt irradiance (LTI), which describe the solar resource available to a flat-plate collector oriented horizontal to the earth’s surface. This approach was used to produce the Typical Meteorological Year Three (“TMY3”) 1991 to 2005 data distributed by NREL as part of the US National Solar Radiation Data Base (NSRDB), which holds solar and meteorolocigal data for 1,454 locations in the US and its territories. This update to the National Solar Radiation Database includes hourly solar data and statistical summaries, including daily statistical files, hourly statistics files and threshold files. While TMY3 was a significant improvement over TMY2 (insolation data from 1961 to 1990 for 237 sites in the US, Guam and Puerto Rico), the measurements are still being refined and the data cleaned for improved spatial, temporal and spectral resolution.
Until the GOES-R series of spacecraft, currently in the development phase with the first satellites scheduled for launch in fiscal year 2015, are operational, major improvements in satellite insolation measurements are not expected. The GOES-R will include advanced sensors specifically designed for measuring insolation at high spatial, temporal and spectral resolution. To improve the data sets based on the currently available measurements, researchers at ASRC, other academic institutions and commercial enterprises are applying new algorithms to the imagery. In some cases, these new techniques improve spatial resolution to three kilometres, and in other cases, improve performance under snow or persistent cloud conditions. The most advanced methods, validated only in the last year, apply infrared channels on the GOES satellite to complement the visible light sensors when the ground is highly reflective, such as in cases of snow cover. Geostellar is blazing new trails in the spatial analysis of remote sensing data to produce models based on measurements from 2006 up to the present moment that are more spatially, temporally and spectrally accurate than those available through other sources.
NREL TMY 3 Data stations in Google Earth
Geostellar applies the most advanced available satellite-generated data sets as a starting point to create accurate, high-resolution insolation models. The first step in this process is to produce a uniform set of values derived from the satellite data through bilinear interpolation and other functions. These algorithms perform linear interpolation first in one direction, and then again in the other direction so that, while each step is linear in values and position, the interpolation as a whole is quadratic in each location. Because satellites derive information from a position high above the earth, the sensor readings assume the earth is flat. If this were the case, our work would be nearly done. However, because topography is highly irregular, we need to apply the satellite-derived data sets to a 10-meter resolution digital elevation map (DEM).
Ground-level and rooftop geomatic insolation models
Geostellar’s application of accurate satellite-derived measurement values to an irregular topography is similar to a technique used in video game and motion picture 3D animation known as ray-tracing. In a movie such as Toy Story, Buzz Lightyear and Woody appear to move rapidly, erratically and realistically through a room with multiple light sources. To produce the effect of light reflecting off various moving metallic, plastic and wood-grained surfaces, where the light sources themselves may also be moving (such as when a lamp is knocked over, or a drape is drawn closed), the animators employ sophisticated software and powerful rendering farms that compute the path of the light as it propagates through the scene and interacts with moving objects. The colors, brightness and shadows generated computationally produce the effects of depth, solidity and flexibility, creating a final motion picture or video game.
In 3D animations, the quality of the product is a function of the resolution. Each generation of animated movie contains more pixels that appear to reflect light independently, until individual hairs and pores are realistically represented. In a similar manner, Geostellar computes the interaction of the sunlight as represented by the aggregated monthly bilinear interpolation of satellite sensor data at three to ten kilometres squared per pixel with a DEM at a resolution of ten meters squared per pixel, then again with a DSM at one meter squared per pixel. In this way, the insolation model is produced by simulating the movement of each square meter of the landscape as it rotates around the Earth’s axis each day and revolves around the Sun each year. With the DEM, Geostellar computes the insolation a half-meter above ground level for utility-scale generation, and with the DSM computes insolation a decimeter over rooftops where commercial and residential generation would typically occur. The DSM computation accounts for shadows from vegetation and rooftop structures such as chimneys and HVAC units.
As each square meter moves on its daily course through the year, accounting for its own individual and unique latitude, longitude, height, slope, aspect and the effect of the wobble of the earth, the Geostellar’s computation engine shoots a ray from that toward the sun at one hour intervals to see if there is an object within eight kilometres blocking the sunlight. An occluding object could be a geological feature such as a hill or mountain, it could be a structure such as a water tower, building or bridge or it could be ground cover such as trees.
When the computation at one hour intervals determines that light has reached the one meter squared spot of earth, it sets the clock backwards a half hour to see if the light had reached the spot at that time. Then it moves forwards or backwards fifteen minutes, seven and a half minutes, etc., halving the units of time, until the computation identifies the precise moment when the spot moves from shadow to sunlight or sunlight to shadow. This can happen several times during the course of a day if there are multiple occlusions that are tall and narrow. The shadow and sunlight computations are then rechecked for each month, as the position of the spot relative to sun changes with the seasons.
At this point, Geostellar has computed the precise moments of each day when the spot is exposed to sunlight, and can take the next step toward computing insolation. To establish insolation for each spot, Geostellar applies the monthly values derived from the satellite measurements using the ASRC methodology, accounting for the value of that sunlight at different times of day and months of the year. The solar irradiance reaching the ground is of a different value when the sun is close to the horizon than when the sun is high in the sky, as it must travel through denser atmosphere in the morning and evening than at noon. The value is also different during each time of the year, based on humidity, cloud cover and other localized atmospheric and climatic factors.
IPL Service Territory Solar Potential
Another important factor in determining the true solar power potential value for a site is the slope of the ground. When the ground is sloped toward the sun there is more usable land, as sunlight collectors can be arrayed up the hill and avoid the shadows of the collectors placed lower on the hill. The insolation values initially determined for a flat surface are amplified as positive values to the extent that the slope is beneficial (toward the sun) or negated to the extent that the slope is detrimental (away from the sun) to ground-cover ratio.
For each moment of the day when the spot of ground is exposed to sunlight, Geostellar computes the solar power potential as kilowatts per meter squared per day, for that particular day. On the first pass of the entire landscape, Geostellar applies 10 meter DEM and monthly averages from satellite measurements to determine the yearly average. The output of this computation is a GeoTIF raster map. Once a particular site or more local area of interest is determined, Geostellar computes more granular values with daily data sets and 1 meter resolution DSM. On specific sites, advanced simulations accounting for the removal of structures and vegetation, or the effects of dynamic structures such as wind turbines, can be performed.
Once an accurate insolation model has been applied to the entire broad landscape, Geostellar processes the resulting raster map to exclude all areas that cannot be developed, then higlights the most promising sites in the remaing area. The following represent typical exclusions:
- Buffers around buildings, roads, rails and other infrastructure
- Parcels under a particular acreage
- Water bodies and wetlands
- Environmentally sensitive and protected zones and habitats
- Agricultural preservation easements
- Hazards such as 100 year floodplains
- Bottom 60% insolation value
- Cultural sites (such as graveyards, historic districts, battlefields)
- Parks and forests
The remaining area, which represents developable land, is then delineated according to absolute and relative insolation values within an area of interest. This process turns the rasterized pixel map that contains essentially analog data (data that can vary continuously between the maximum and minimum values contained in the map) into a discrete data set with quantum identities. For example, input values of 0.234, 12.5252, 13.231, 5.555 would, with the appropriate contour scheme, produce bands of 0, 2, 2, 1. Up to 256 discrete bands, which refactor insolation as polygons, are supported.
To generate sites, Geostellar processes the raw insolation values of each 10 meter or one meter pixel across the landscape into polygons for all continuous band index values. At the same time, Geostellar accumulates statistics for each polygonal site that are applied in the final stage of geomatic computation. The output of this the site generation subsystem is a shape file and tabular dataset describing both absolute and relative insolation values for each uniform area across the landscape.