There is something odd going on inside the U.S. Department of Energy (DOE). It seems that the department offers a solar calculator that miscalculates. The DOE calls it "PVWatts," but it does not calculate energy production worth a hoot. Any misinformation that PVWatts generates could hurt manufacturers and developers such as SunPower (SPWR), First Solar (FSLR) and SolarCity (SCTY).
PVWatts was developed by the DOE's National Renewable Energy Laboratory's (NREL) Electricity, Resources and Building Systems Integration Center. According to the lab's website, "PVWatts calculator determines the energy production of grid-connected photovoltaic energy systems throughout the world. It allows homeowners, installers, manufacturers and researchers to easily develop estimates of the performance of hypothetical PV installations."
NREL further explains, "PVWatts calculator works by creating hour-by-hour performance simulations that provide estimated monthly and annual energy production in kilowatts. Using typical meteorological year weather data for the selected location, PVWatts determines the solar radiation incident of the PV array and the PV cell temperature for each hour of the year. The AC energy for each hour is calculated by multiplying the DC energy by the overall DC-to-AC derate factor and adjusting for inverter efficiency as a function of load. Hourly values of AC energy are then summed to calculate monthly and annual AC energy production."
PVWatts will provide users with a summary table of results. Just looking at the summary table will not reveal the problem. Analysts must look at years' worth of hourly data to see the model's propensity to degrade output at precisely the wrong hour. That hour is typically during the peak production period of the day, when market prices are also near their peak. Take a look at this graph:
First, notice that Worcester, Mass., is projected to outperform Richmond, Va., or Miami, Fla., for the same days. There is something wrong here. Perhaps humidity and other environmental factors may offer an explanation, but NREL's contour maps suggest that Florida should generally outperform Massachusetts.
Second, notice the daily hiccup, which takes place during the peak hour of most days. If all PVWatt data were to be analyzed, it would be evident that a hiccup occurs at about the same hour on most days for most geographic locations. In contrast, it is rare to see a hiccup during the early or late production hours.
What is the big deal? The curves are used to determine an investor's top-line revenue and bottom-line earnings. If the curves are wrong, investors' expectations will be wrong. It appears in PVWatts case; the DOE's curves are incorrect and should not be used as the definitive basis for investment decisions.
It is worse than most financial analysts think. Solar panels' output changes, depending on the sun's hourly location. When the sun is rising and setting, a typical panel produces only 20% of its nameplate rating. Only when the sun reaches its peak does the panel approach its nameplate rating. That is also when higher market prices for power appear. Should NREL alter peak-hour production data, that would significantly alter the amount of revenue that owners and investors can expect from new solar facilities.
Admittedly, it is almost impossible for anyone to develop a one-size-fits-all model. For example, weather conditions are unknowable in advance, but they will definitely alter the solar facility's performance.
To be fair, DOE has published disclaimers. It never claimed that PVWatts was accurate. NREL never suggested that bankers should rely on PVWatts to finance solar projects. Nevertheless, many analysts ignore the DOE's warnings and rely on PVWatts for most of their due diligence.
If the Obama administration and DOE want to promote solar power, they might reconsider their PVWatts model. They could make the model more useful and less prone to misunderstanding if they allowed the user to control more information. For example, the National Oceanic and Atmospheric Administration (NOAA) has enormous amounts of data, including sun positioning and solar data. NOAA's National Weather Service (NWS) has additional weather histories covering many geographic locations across the country. The DOE should give users ready access to that data and allow them to make appropriate adjustments.
In the interim, if investors are attempting to investigate a solar investment, they should avoid relying on DOE's PVWatts program. At best, PVWatts is an order-of-magnitude device that can potentially produce significantly troublesome errors.
