The primary issue for the user of any model is whether it is appropriate for the application or problem under investigation. Does the user have some measured data to evaluate and compare with a model result? Does the user need to evaluate one or more particular collector configurations or designs (Concentrator or flat-plate collector? Optimized or constrained tilt or azimuth aspect for a collector?)? Is a clear sky “maximal envelope” of resources sufficient? Is a realistic time series of data needed? Are the actual input parameters required for a model available? If not, is a suitable surrogate for missing input parameters available? As mentioned many times here and in the individual chapters, the overriding concerns regarding all of the models discussed (and models in general) are accuracy and uncertainty. All of these questions lead us to consider the following needs for future research:
1. Availability and quality of measured solar radiation data must be improved. This includes (a) better radiometer design for more accurate data; (b) more widely distributed, geographically complete measurement stations; (c) higher time resolution data as often 1-s data are needed for studying transient behavior of very large photovoltaic arrays; and (d) more public, easily accessible data (Web portals, data applications).
2. More and better model input data are needed. This includes the rather esoteric aerosol optical depth (AOD) and precipitable water vapor (PW) data required for most models.
3. Higher time resolution input data are required. AOD and PW data may sometimes be obtained for regions with similar climates but usually on a monthly or annual mean basis. Daily and hourly data will provide better, more natural time series data for more accurate system design and performance studies.
Regarding the models themselves, it is clear that little improvement has occurred since the mid-1990s. Most models discussed here are comparable in their performance. The need for more accurate models, or more accurate model uncertainties, requires the following:
1. Validation of all the broadband models described here (all were based on measured hourly average data) for various higher time resolution data, such as 10-min, 5-min, 1-min, and even 1-s “instantaneous” data.
2. Simple but accurate means of “extracting” high time resolution or small time step data from “summarized” data such as hourly average, daily average, or monthly average data (whether measured or modeled).
3. Development of a simpler, but more accurate clear sky model using less – esoteric (AOD, PW) input parameters that are easily (preferably universally) available.
4. Validation of the Perez anisotropic tilt conversion model over a wider range of climates and tilt configurations. The original model was only validated
for vertical tilts (90°) in the cardinal directions and 30° south-facing tilts. Refinements of the e, A, and Fij parameters would improve the model.
5. Evaluation of the Maxwell DISC (direct insolation simulation code) GHI – to-DNI conversion model equations over a much wider range of latitudes, including a wider range of (low) zenith angles. The model is known to underpredict DNI at low zenith angles because of the limited data set (continental United States) used to develop the model equations.
6. Better, more accurate correlation, with smaller random error, relation between GHI and DHI irradiances. This supplements and complements an improved DISC model.
7. Better cloud cover modifiers, to somehow account for cloud spatial distributions with respect to the sun, either through (easily accessible, free) sky dome imagery, satellite imagery, or cloud forecast/hindcast data. This will result in better cloud transient modeling of solar radiation available to large – area (several-square-kilometer) systems.
As of 2012, many of these are current topics of research, especially for solar resource forecasting and future solar radiation database updates. Obtaining the financial, intellectual, and material (measurements) resources needed to meet these needs in difficult, or even less-stressful, economic times is a daunting task. But, given enough time, money, and computer power—at least that is the hope. I have the fervent belief, for the sake of future generations and the sustainability of the environment for them, that these goals, as well as the existing tools described in this book, can be powerful tools for success in fostering renewable energy systems of the future.