By Gwen Bender, product manager for solar assessment services at Vaisala
Weather is the greatest source of performance variability for solar projects. Unfortunately you can’t look into a crystal ball and be 100% accurate about the weather at a site all the time. Instead, solar developers must focus on reducing their level of uncertainty as much as possible.
Predicting the weather (solar irradiance, rain, wind, snow and more) is essential to estimating system production and profits, and it’s increasingly critical that solar developers know how to use available weather data, along with weather measurement equipment when necessary.
More data providers
When it comes to reliable data, thinking long-term is key. In the not-so-distant past, options for long-term climate data—looking at weather patterns over years or decades—were limited to those from public sources such as NREL and NASA. These days, there are multiple high-quality, long-term datasets available from additional public and private sources.
To increase accuracy, today’s providers often use data from multiple irradiance estimation models. For example, Vaisala compares five different in-house models when assessing solar projects and has recently released an online tool to allow clients to do the same. A subscription to its online Solar Time Series Tool allows developers to view a resource map, see long-term weather trends from the multiple models and download time series data directly from the tool. Services like this help project developers evaluate and access better resource data to determine more accurate energy estimates.
Figure 1: Vaisala’s Solar Time Series Tool
Consider the source!
Most providers’ datasets use the same base satellite information but vary in their inputs and methods of irradiance calculation. With more options it is increasingly important to understand the dataset version being used—not just the vendor. Just like with software programs, version numbers change as data is improved. Each version will have different outputs and also different levels of model uncertainty associated with each version.
Typical meteorological year (TMY) data remains the industry standard, but it is easier than ever to get long-term datasets or more exotic data options. An example would be P90 year file which models the “worst case” production year of your plant in comparison to your “typical” P50 year.
The software programs developers use to translate resource data into potential energy production are correspondingly becoming more sophisticated in accepting inputs. Until recently the major platforms could only accept and process a single year of resource data, the expectation being a TMY file. Now you can process multiple years of data at a time or input data from your ground station.
More datasets = more confidence
Developers can look at multiple datasets together to become more confident about the accuracy of their resource assessments. If the data sets provide similar estimates, then the developer can be more certain about a particular site’s weather. However, if the datasets are much different from each other, the developer can be less certain about a site.
Compare it with someone else’s work
Another way to look at datasets and reduce uncertainty is to compare the data to an independent “truth.” An example would be ground station observations, derived either privately or from nearby public weather stations. This can help developers determine which dataset is showing the best correlations for their project site.
Figure 2: Even a short period of observations from the World Bank at this site in Pakistan can be informative in choosing which long-term record to use.
When to get out the pyranometer
All models have some degree of uncertainty. Again, no one has a crystal ball to determine the weather, but developers do have access to the next best thing: pyranometers! While irradiance observations themselves are not perfect, they can be extremely useful in helping to make decisions about what dataset to use, especially in datasets that show more deviation and uncertainty.
Vaisala also offers on-site weather measurement equipment in its “integrated observation station package.” Developers can install this equipment onsite to help measure wind speed, temperature of both the air and the solar panels and precipitation levels. This information can be incorporated into other datasets and help guide developers to better racking choices and understand how factors like rain or snow can affect system production.
It all pays off
Though steps to reduce resource uncertainty is an added expense, they enable developers to more accurately estimate production and profits—which is important to banks and other stakeholders. Research from pyranometer manufacturer Kipp & Zonen has determined that developers save $20,000 on average through improved financial terms for every 1% of resource uncertainty reduced. The exact amount will of course vary depending on project circumstances.
Using long-term resource data can help developers be 7 to 10% more accurate about the resources they have on a project, which then helps them be more accurate with production estimates and profits. Incorporating onsite observation equipment can ensure even greater accuracy.
Maybe it’s time to change it up
If you have always used the same resource data sources in your solar projects, now might be the time to change things up. It’s not only easier to evaluate multiple resource data sources now, but there is also some danger in having your entire portfolio based on the same source data.
If, for example, you have used the same resource data source in the same version of PVsyst for years, you run the risk of systematic bias affecting your entire portfolio. By evaluating resource data sources on a project basis, you can reduce uncertainty at that project and across your portfolio. Diligent resource assessment can help generate long-term cost savings and enhance the overall bankability of your projects.