Guest Post by Jani-Petri Martikainen. Jani-Petri is a theoretical physicist doing fundamental research in the field of ultracold quantum gases. Most of his current research activities are computational and involve bosonic or fermionic atoms in optical lattices. He has a lively interest on environmental, climate, and energy issues. He runs the blog PassiiviIdentiteetti, which is mostly written in Finnish.
Jani’s previous post, Geographical wind smoothing, supergrids and energy storage, focused on distributed wind alone. In this follow-up, he turns his attention to solar combined with wind.
Earlier, I wrote on how crucially an unreliable sources of power such as wind depend on fossil fuels. Based on real world production data from around the world, I noted that even with massively distributed production wind power is very variable and necessitates a reliable backup power source (typically from fossil fuels) which must be able to produce essentially all the power society consumes. A way around this problem would be a massive energy storage, but I found the size of the required storage to be unreasonably large.
One typical response to findings such as these, is to brush them aside by claiming that even if true, the results will not matter since we will have many different renewable energy sources acting together (as if there is some “harmony” in two essentially random signals). Most importantly quite a few people base their vision of future energy production on a mixture of wind and solar power. For this reason I felt the need to return to this problem so that also solar power is considered. Unfortunately, I have yet to find a good source for real world production data for solar power. The best I have come up with are images (typically of the daily production), but raw data is better hidden.
However, since solar power (without storage) production is proportional to insolation we can use meteorological data as a reasonable starting point. US has a National solar radiation database which has large collection of insolation modelling data around USA. From this data they have also formed a “typical meteorological year 3 (TMY3)” datasets. (There are some quirks in the construction of TMY3 that I frown upon. For example, after El Chichón and Mount Pinatubo eruptions insolation was reduced, but these periods were apparently excluded from the TMY3 as atypical. Of course they were atypical, but they are still things that do happen and whose effects must be considered. However, I suspect that the effect due to eruptions was still minor in US.) As my insolation data I take the average of TMY3 data from six different class I sites (class I has the best data) in three different states: Prescott Love and Tucson Airport in Arizona, Arcata Airport and Fresno Yosemite Airport in California, and Denver Airport and Limon in Colorado. These sites have an average latitude similar to southern Spain. (Why did I choose these sites? Well, being lazy I started from the entries listed in alphabetical order by states and picked the first southern states I encountered.)
Somewhat annoyingly only hourly data is provided. We know from BNC among others that solar power (especially PV) can have large swings on shorter timescales. Therefore, this limitation may have important consequences. Nevertheless, let us ignore the torpedoes with an understanding that the solar power we talk about here is such that sufficient storage has been already implemented to smooth out hourly variation in production. So keep in mind, that the starting assumptions for solar production have a bias towards the optimistic side. Since the production data for wind power is given every 5 minutes I will linearly interpolate the solar insolation data to deduce the production of solar power every 5 minutes (link to the data here). As in the earlier study the data corresponds to one year starting July the 1st. and the consumption data corresponds to the Bonneville Power Authority load with a possible scale factors to suit my needs.