The three-hour dilemma; solar PV and peak demand

Installation of residential solar PV panels is ramping up globally; in the USA, installations more than doubled in the 2nd quarter of 2012 with a nominal capacity of 5.16GW across the country now [1]. Australia, however, is comparatively far ahead of the USA in that regards, with over 2GW of installed capacity despite 1/14th of the population. And it is here that we start to see both the potential and the dilemmas play out.

In sunny countries, electricity production with solar PV panels is, at face value, ideal – the peak demand is typically when it’s hot, i.e. when it’s sunny, i.e. it aligns perfectly with the solar PV production peak.

Well almost perfectly.

And it turns out that that “almost” bit has the potential to cause a lot of grief in the grid. In South Australia, the fast ramp-up of solar PV generation capacity has served to shave off as much as 8% of the daytime & afternoon peak demand (image and figures via [2]):

That’s awesome! Right? Maybe, but it’s not only awesome – it’s also a huge headache for the power companies, and thus you the customer as well. As you can see from the graph, demand has been lowered until about about 6pm, which is when the both the sun starts to set (decreasing production from solar PV) and when people return home from work – and crank up their air conditioners, increasing demand. This is not a theoretical exercise; it is already estimated that for every $1,000 people invest in an air conditioner in Australia these days, another $3,000 needs to be spent on upgrading the grid to cope with the additional load [3].

So why exactly is this a problem? For one, the grid (production, transmission, everything) needs to be dimensioned for the peak demand or it will fail. Second, electricity companies have long made a bulk of their profit from the peak hours – peak hours which are now much shorter. Contrary to what one might think, this is not evil of them – wholesale prices are naturally highest when the demand is highest. Take away those peak hours and you take away the revenues (and just as a reminder, a business without revenues will fail, taking the service it provides with it).

Because the peak hasn’t been lowered – only shortened – power companies are under pressure to raise prices. This is of course only one reason for the price hikes, but an important one – and one reason why Australia is cutting feed-in-tariffs left, right and center; they will be only 8c/kWh in Victoria next year, compared to a rather generous 60c/kWh a couple of years ago.

What emerges from this is that one of the most important questions to answer in the energy sector today (if we work under the assumption that solar PV is a good thing) is how to shift the peak demand and peak production by just a few hours to better coincide? Alternatively, we need to lower the now-very-short peak period to a permanently lower level. This is not just a problem, but it’s a multi-billion-dollar opportunity.

I can see a number of ways to bring that short but high remaining peak usage down or shift the solar PV production peak to more closely align with the demand peak; there are both centralized and distributed solutions and both technological and societal / behavioral solutions. I’ve mapped some of these ideas I have onto this table:

Some of these are self-explanatory, some may require some elaboration;

  • By far the easiest and arguably the best – easiest in the sense it wouldn’t require any large-scale behavioral, or even grid, changes – would be distributed energy storage. If we go with the centralized storage (e.g. molten salt storage for CSP plants) it still means the grid needs to be dimensioned according to the peak demand. If, however, the storage is distributed and co-located with the consumption, it reduces the overall peak demand and allows maintaining a more modest grid infrastructure, thus lowering costs. Today, however, this means expensive and somewhat bulky batteries. In the future, maybe things like hydrogen fuel cells could be used, but either way achieving this in the near term future would be expensive.

  • Pricing is obviously one strategy for forcing (“encouraging”) behavioral change, although it can be seen to unfairly target poorer customers. Even with the peak wholesale price capped at $12,500/MWh here (or $12.5/kWh), there’s obviously an option to charge consumers obscene amounts at peak – OR offer them incentives to achieve the same. If people were forced to pay obscene amounts for the peak usage, they would adapt usage – but many begrudgingly and under duress, so this is not the best possible solution by any means. But what if you were paid to shift your demand?

  • Many of the solutions require massive investments; however, there is one standout thing that could be accomplished relatively cheaply: D5. With only a little bit of added smarts, the house could be pre-cooled before the solar PV production starts declining – this would in turn lower the peak demand as the units wouldn’t have to work so hard at the worst possible time.

  • On the “centralized behavioral” section, significant portions of the workforce could go home earlier or later – maybe, in the longer term, in step with adopting the 21-hour work week proposed by the New Economics Foundation [4]

  • The solution with the lowest environmental impact and no technology required would of course be to have the people outdoors; it’s a curious problem to have, having to find ways to incentivise people to go to the beach on a hot evening rather than sit in front of the TV in an air-conditioned indoor space.. A sad sign of our times?

  • Some other tweaks could also help; things like orienting the solar panels towards the evening sun so that the production peak happens later, albeit at the cost of overall efficiency. A better improvement would be sun-tracking panels, solutions for which already exist – but are expensive.

I believe that through some combination of these we could have a material impact and significantly lower the demand variability on the grid. What other (partial) solutions can you add to the list?


This entry was posted in Energy and tagged , , , , , , , , , . Bookmark the permalink.

2 Responses to The three-hour dilemma; solar PV and peak demand

  1. Andrew says:

    Interesting! To what extent does spreading the grid across a significant chunk of longitude (e.g. multiple timezones) help?

  2. sim says:

    Andrew, that’s an interesting idea. The answer probably is that we don’t know, because it hasn’t been tried – and in Australia, cannot be tried due to grid structure – to a sufficient extent. One major issue would be transmission losses which are 3-5% per 1,000km.

Leave a Reply

Your email address will not be published. Required fields are marked *