The Impact of Weather Events on Solar Energy Generation in Canberra

In a wonderful union of my two favourite scientific subjects, I've been working with ANU student Sonya Wellby to identify the weather events that results in large scale, collective PV ramp events in the Canberra region.

And our findings are fun and exciting!  We'll be presenting them today (13 November 2014) to policy members, ActewAGL (the local distributor), fellow scientists and the general public.  More on the venue, timing, etc can be found here

This blog post is meant to play a supporting role to that presentation, so that those of you who are interested in looking at this subject a bit more in-depth, may do so.  We'll be producing a scientific publication, which I will inform you about in early 2015, that will get into the nitty-gritty of what we've done.  I also will post video/slides/audio from today's presentation in another blog post for your reference.

let's get down to business..

There are several key points that help introduce the subject matter:

  1. Solar PV systems have a power output that varies according to the available solar radiation
  2. The supply of solar radiation is often interrupted by clouds
  3. Cloud cover is driven by (and is a part of) meteorological events
  4. With tens of thousand of PV systems installed in relatively dense geographic regions in Australia (e.g. capital cities) - it is possible for a cloud event to cover all of them at once
  5. Weather events are generally repeatable/follow patterns
  6. We decided to identify the weather events that create sudden, broad-scale changes in all the solar PV systems in Canberra

This exercise in identifying and categorising weather events is a first step in learning how much energy generation from PV systems in a region changes during these events, and then predicting the events ahead of time so that the supply-demand of electricity can be appropriately managed. 

A negative ramp event; black line is observed power output from 200+ systems, dotted blue line is clear-sky output.

Tropical, convective clouds move over Canberra from the NE (marked by purpled dot)

significant, collective ramp events

There were 12,000+ solar PV systems installed in Canberra by the end of 2012.  Through cooperation with ActewAGL, our local distributor, we've been provide the postcode and rated capacity of these installations.

We've also been able to collect data from 200+ sites in Canberra which have reported their power output publicly, drawing from a variety of available web servers.  

Using these 200+ sites, Sonya searched through images of their collective power output (meaning all 200+ sites added together) searching for times when the collective power output from these PV systems changed very suddenly - this is termed a ramp event.

what's a ramp event?

I've addressed this topic a few other times on this webpage (see here or here), but I'm a nice guy, so I'll do it again.  A ramp event is a term used to describe a situation where the power output from a  PV system experiences a large change very suddenly.  A positive ramp event implies the power output changes from a low value to a high one, and a negative ramp event is the opposite.

In our case, Sonya was searching for ramp events where all the PV systems experienced a ramp at approximately the same time (we term these 'collective ramp events').  These types of events are important, because they indicate that all the PV systems in the region (12,000+) experience the same change - and that means big changes in the supply-demand balance of the electrical grid.

who cares? 

This type of event doesn't matter unless there is a high penetration of solar PV systems in a given region, meaning that more than 20% of the electricity is being supplied by the collective generation of all the installed systems.  It just so happens that in the ACT, we have a 90% Renewable Energy target (one of the best in the world!).  As a result,  we are installing several large solar farms (20MW plant at Royalla just opened a few months ago) and the uptake of rooftop PV continues to steadily increase.

So if there is anywhere in the world where solar forecasts matter, it's likely that Canberra is it.  

Now I can only say 'likely' because this whole electricity grid thing is really complicated and we simply don't have enough information about the location/arrangement of the PV systems in Canberra to make a final conclusion.  But that's part of what today's presentation is about - working with ActewAGL to make sure the continued integration of heaps of solar in the Territory can continue.

what did we find out?

Over a two year period, Sonya identified 19 positive and 16 negative significant, collective PV ramp events in the ACT.  She then categorised the weather events that lead to these events, which are presented in the following table:

This is a great news, because all of these weather events are things that our weather models and forecasters at the BoM are able to predict with lead-times of several hours.  With that type of information, any time we see that the sun won't shine - we can deliver electrical energy from alternative sources.  On the flip-side, when the clouds are about to clear out - we can stop purchasing electrical energy we won't need.  It's a beautiful thing!

let's take a look at some cool stuff

Enough talk, let's take a look at some of these weather events in action.  The following videos come from my research efforts, using my KPV methodology (link to publication) to simulate all the performance of all of the PV system in the ACT.

negative ramp event - 30 March 2014

In the first video, we have a negative ramp event that results from a thunderstorm event - watch how the clouds move in from the SW to obscure all the PV generators in less than 1 hour!  A 20MW+ change in total power production (I estimate that number increases to 60MW with the addition of 2013/2014 PV installations, including the Royalla solar farm).

Radar at 1:38PM

Radar at 2:38PM

positive ramp event - 30 March 2014

The next video shows a positive ramp event, where fog and low cloud burn off over the course of ~90minutes.  This is a positive feedback cycle, where the more the fog/clouds clear, the more sunlight makes it to the surface and the more rapidly mixing out of the fog/cloud occurs.  It's pretty amazing to watch.  Notice how the fog/low cloud sticks around Lake Burley Griffin/Ginninderra the longest.  Meteorology is awesome!

where to from here?

First, I hope this will encourage cooperation between policy makers, solar farm operators, ActewAGL, the BoM and my research program.  The open sharing of information will allow for world-class research to be undertaken here in the ACT.

Second, we'll use this research and the resulting publication as a launching point for a grant application to ARENA to repeat the same exercise in all high-concentration regions of solar generators (e.g. capital cities).  I've already line-up some students for the initial work.

Third, continue the proliferation of open PV data sharing, particularly from members of the public, via projects such as our machine-learning forecasting efforts, and across research groups. All of my research code will soon be available via an R-package, which I hope will encourage future research like this everywhere its needed.

in closing...

Here's a Brief of Research Findings prepared by Sonya Wellby, who completed the identification of these weather events via a Special Topics Course (ENVS3016) in the Fenner School of Environment and Society at The Australian National University.  She's the most capable Australian student I've ever met - so watch out for her in the future, she's going to make things happen!

That's all for now - stay tuned for the presentation 

Questions, Comments?

please connect using the social icons at top right and bottom of this page.