Renewables Sharpe Ratio - NASA found a wind park in Tasmania

Showing first results of this post to a co-worker of mine, he said: “You did a great job turning Australia into a pink porcupine”. Yes, and I will show you, that every brightly colored spine is a good wind spot.

Obviously from one’s gut feeling it should be an attractive location for installing a lot of PV farms and due to the huge coastline there should be great wind spots as well. Remembering the balancing effects between wind and solar Australia could be a perfect playground. Not to forget, that the biggest Power to Gas plant will be built in Australia. Looking back at my second post, this country could proof itself as a suitable precedent for a balanced renewables power plant with a gas storage solution comparable to the Kennedy Energy Park project. In theory all of this sounds nice, but how can we find good locations for potential power plants? We will try to answer this question with massive data analysis and adapting a simple ratio to get something like a scoring system for promising wind farm locations in Australia.

For carrying out massive data analysis, surely a lot of data is needed. We collected all kinds of meteorological data from NASA and GSODR. Hence, we have a giant database of all kinds of weather data at hand, which we can use to gain insights. Thanks to Adam Sparks for giving me the hint to this wonderful and free data sources.

Analysis approach

Having acquired wind speed data from different sources, there is the simple Betz’s Law, which allows us to transform the wind data into potential production volume of a windmill. Again, we will use the model from our last post, namely the Gamma distribution, to deduce statistical results from the given wind speed data, remembering that this distribution is a good choice in regard to tail behaviour. In order to measure the goodness of potential locations, we adapt the Sharpe Ratio to our wind production setting. Let us define:

$$\begin{align*} S &:= \frac{E[X]}{\sqrt(Var(X))} \end{align*}$$

where $E[X]$ is the expected wind production and $Var(X)$ represents its variance. Latter is derived from the estimated Gamma distribution parameters, expected production is set as the mode of the gamma distribution. General setup of the algorithm is quite straight-forward. Looping through longitude and latitude, we will fit the gamma distribution on a de-asonalised, winsorized logarithm of the three year’s daily power production history estimated by Betz’s law and calibrated with moment’s matching method. For each grid point we will calculate the above defined sharpe ratio. The higher this ratio is, the better is the wind location. In other words, there should be high production volume with low variability in production returns.

We split Australia into a fine longitude and latitude grid. For each grid point we carry out the analysis above based on different data sets. In order to present results in a flashy way, I used the wonderful mapdeck package. Interpretation is quite straight-forward and less technical. Spots, where columns are brightly shining, high and dense represent good potential wind farm locations.

Flight through data

Looking at this small satellite flight, we discover bigger areas on and off-shore, where a stable wind is blowing. Central Australia from west to east throughout Alice Springs seems to be a miserable place for wind power plants. Off-shore regions south and north of Australia are more promising and especially Tasmania could be an attractive location. Generally, the fine gridded NASA data gives us a overall impression of good meteorological regions, but even if we zoom in with our satellite, it is hard to mine specific spots for wind farms.

To put another layer into this analysis and in order to concentrate on specific spots, we further investigate wind farm locations with GSOD data set.

Surface weather stations in Australia

This data is provided by weather stations on Australian soil. We will carry out exactly the same statistical analysis as with the NASA data, but now on fixed surface weather stations. Combining both should give us not only global overview, but also more spot related potential wind farm locations.

Evidently lower east coast is a devastating good location, which is not only because of good levels of sharpe ratio. Certainly, the density of weather stations has something to do with the more optimistic looking picture at Perth, Adelaide to Canberra. Besides a lot of wind farm have been already built in Victory and New South Wales. So, it is not really a secret, that these areas are reliable wind farm locations. Let us harvest areas, where not so many weather stations are in place. By doing so, we for example arrive at remote locations like Springsure and Mount Isa in Queensland.

But of what use is a good wind farm location on Coconut Island? Evidently we need some more infrastructure in order to get the produced power to the customer. Hence, we validate our insights with the maps of the Australian energy market operator (AEMO). Combination of NASA, GSDOR and grid data (AEMO) should tell us at least to a certain amount spots, where it could be promising to build a plant.

Comparison of AEMO data with sharpe ratio map

Obviously grid availability is rather imbalanced in Australia. South-Eastern Australia with its high sharpe ratios has developed grid connections. Inner Queensland or Southern Australia have only limited capabilities. Nevertheless there exist plans to enhance the grid capability, because also AEMO has recently discovered that the best wind locations in Australia are very remote.

Summary

Obviously New South Wales and Tasmania look like promising regions. New South Wales has already seen some development, when it comes to wind sites. On the opposite Queensland would be more interesting, as there aren’t so many wind farms built yet. For example looking at our sharpe ratio map, we would perceive the area around Muttaburra as a promising region. Unfortunately, as a frustrated developer in a recent newspaper harshly said, Queensland is dead to us. The sunshine state could surely install a lot of solar plants. Accompanying these plants with wind would generate good balancing effects, which is an alternative to any gray energy. However relatively new coal power plants were built recently, so seemingly there is a real competition between renewable energy and gray energy.

Again looking at the map, also Tasmania seems to be a promising region. Short research shows, that only 2 operational wind farms exist in Tasmania. Most of energy here is delivered by hydro power plants, which could also work like a battery. Moreover, there exists an undersea grid line to the mainland, which could be used for further sales purposes.

Sources

Data is obtained from the NASA Langley Research Center POWER Project funded through the NASA Earth Science Directorate Applied Science Program. It goes without saying, that with the help of NASA we could carry out lots of analysis on wind, solar and hydro in every country we wish.