In conjunction with this larger post (including animation) looking into what happened in the Queensland region on “sizzling Saturday” (14th January 2017), I’ve powered up NEM-Review v6 to create the following chart:
It’s not 100% accurate (as there are other factors to consider) but adding Rooftop PV (AEMO estimate*) to the Scheduled Demand for the region provides, in aggregate, a view of approximately what total consumption has been across the three days shown (Friday 13th, Saturday 14th and Sunday 15th).
* with respect to the AEMO estimates, see my prior note about the opacity of small-scale PV. Thankfully with AEMO data we’re able to start publishing about these insights.
Ten years ago, “Consumption” and “Scheduled Demand” might have been pretty much the same thing – but increasingly the two are diverging, and both need to be kept in mind to understand where the electricity system is headed. This chart is fairly typical of what we’re seeing – and is not unique to Queensland, either. What it shows is two related things:
#1) The batch of rooftop solar PV currently installed has definitely reduced peak demand
This is very apparent in the chart above – which is just a random example. On Friday 13th January, for instance, we can clearly see that the demand would have peaked above 9,500MW had it not been for rooftop solar PV. Similarly, on Saturday 14th January, the demand would have peaked above 9,000MW – and so on…
#2) … however more of the same won’t deliver more benefit (and is likely to bring its own challenges)
As per the second note on the chart, however, because the time of peak demand has shifted to late afternoon (when rooftop solar PV injections are waning) the addition of more solar panels won’t deliver any more benefit to the market in terms of reduction of peak demand.
Indeed, because of the “duck curve” phenomenon we’ve spoken about previously on WattClarity, it will come with its own challenges.
Don’t take this as a reason to stop building solar PV – however please do:
i. Stop promoting it on the basis that more will reduce peak demand more; and
ii. Understand some of the challenges that come with it.
** UPDATE at 17:30
Two of our valued readers has let me know that the above is not that clear – for which I apologise. Both readers had two particularly useful points to make – hence the update here:
#1) Assuming north facing
When I make the comment that “more solar PV won’t reduce peak demand more”, what was I left unstated was the “… if it’s more of the same”. With this I particularly mean the current incentive structure geared towards a kWh harvest, which provides the (quite logical) motivation for people to install PV on north-facing rooftops. As seen in the chart above, this has had a great effect on reducing demand in the middle of the day – however that’s not where the peak demand is any more (see point 2).
If the incentive were to change, say, to promote west-facing panels (for at least a reasonable proportion of new installs) then we would see the afternoon peak reduce more – however it would be at a “cost” to the owner of a lower kWh harvest.
Hence the question why would they do this? Not too many are purely altruistic.
#2) It varies by location
Another telling comment was made that the details are buried in the distribution system – down at zone substation level, or even lower at the feeders.
We’re working on data sets down at that level that might help with that puzzle…
Hi Paul,
At some point in the future, when you may be looking for something to do, it would be interesting to see what the $ impact on the market was due to the solar output. In other words, quantifying the avoided cumulative cost by the depressing nature PV has on the spot price (merit-order). Conservatively, this may assume the existing spot price, i.e. spot price during solar output. This would be instead of trying to estimate a new marginal generator and raising spot price.
With the current availability of data it is quite impossible to determine the actual total output from small scale solar systems. The APVI data is based on such a small number of the installed systems, and even that sample shows gross distortions in system output, that expansion of the data to provide a whole state estimate is almost ludicrous.
Even the systems that are reported through PVOutput.org are more likely to be recently installed (more likely to have WiFi outputs) and from systems installed by those who are more likely to follow regular maintenance routines. So the base data is likely to be heavily skewed to the “better” performing systems.
I would be very keen to see calculated confidence intervals for the APVI data, but my guess is that they would be so wide as to make the estimate of “peak” output from solar systems very difficult.
As the AEMO data is quite similar it is likely that they follow a similar procedure as the APVI data; i.e. relying on a very small number of systems actually reporting output.
I therefore suspect that picking peak usage by relying on this data is very dangerous. But that’s not really the point. There are all sorts of methods for controlling the actual peak demand on any one day, uncontrolled solar input being just one, but there are also many methods which would provide some control over the peak.
The real issue for consumers, industrial and commercial, is the price and reliability. From your graphs it is clear that the behaviour of the small scale solar units causes a narrowing of the peak demand on conventional generators. To maintain reliability we still need conventional generators capable of covering the actual peak (in the event that solar output is low). So we have not saved any capital but have pushed down the operating hours over which this capital can be recovered and so sending up the price. We can see that in the peak pricing as the solar systems start to fade.
So small scale solar system are not really helping the peak demand issue, they may in fact be making it worse. Irrespective of whether they actually reduce the peak or not.
On the possibility of west facing solar panels – it is very interesting to see the additional output from single and dual tracking solar systems. But all of this, including batteries, just keeps making solar a more expensive, rather than a more competitive power supply option.
Michael, the diseconomy you have pointed out of running base-load power stations over a narrow peak period underlines the unsuitability of coal or nuclear baseload power as a backup or complement to renewables. What we need is not baseload but dispatchable generators, like gas, hydro or solar thermal with heat storage. From this perspective it is extraordinary that the solar thermal option is routinely ignored in discussions like this and, of course, by government..
Although an enthusiast for PV solar, I can’t see the point of installing large PV arrays over flat, agriculturally useful land. There are very large areas of commercial shopping centre, warehouse and factory roofs as well as carparks and walkways near existing powerline infrastructure where PV panelled roofs would provide shade and power at zero land cost.
All investment in land-covering solar should be focussed on solar thermal with energy storage.
Hi Paul,
the spot market electricity price in Qld seems to be spiking more and more in late Jan/early Feb. I can’t recall ever seeing such extensive spikes. What is going on ? and what can we expect this to do for the next round of power price rises from the retailers ?
I have a 3.8kW PV array on frames to reorient them N on my WNW roof slope. The reason for this is to locally power the washing machine in the morning on days when the solar dryer (clothesline) will then dry the wash that afternoon. By such load management our usually 2-person household pays nothing for electricity, only interest on investment. Energy exported at 25c/kWh for the first hour and 8c/kWhr thereafter pays not only for power drawn (at 25c/kWhr) from the grid at night and cloudy weather but pays for the line rental as well. The solar hot water helps considerably of course.
It does not matter that the peak output does not coincide with peak airconditioning demand because we don’t have an air conditioner. Our Hardiplank/brick/Colorbond house is well insulated and we open all the windows at night (in summer) and close them and curtains during the day. But even if we did have an air conditioner we, like everybody else, could effectively match our demand to the PV output peak using a timer on the air conditioner. With a sensible pricing scheme utilities could make it worthwhile for everybody to run their air conditioner during the N facing PV peak and shut it off as demand rose for the evening cooking demand peak. Heat and “coolth” represent storable energy and if the house is cool when people arrive home from work it can stay cool through the evening.
Storage batteries and pumped hydro would be useful for other load mismatches but we don’t need them for hot days. Thermal mass does the trick.