The Energy Collective Group

This group brings together the best thinkers on energy and climate. Join us for smart, insightful posts and conversations about where the energy industry is and where it is going.

9,751 Subscribers

Post

And the Winner Is: Distributed Generation

Not a day goes by when I don’t learn something new about what solar energy can do and frankly, how many falsehoods exist about “the problems” that it theoretically causes.

I’m no conspiracy theorist but I can’t say I blame them for being suspicious. But what I love most is when facts and data bust the myths being spouted and today, we got an extremely cool new set of truths. Arguably, they could have quite profound global ramifications and this project was conducted right here in Australia.

One of the most common myths we hear is that high solar penetration rates are devastating for the network. High penetration solar can cause changes in voltage and frequency and well, apparently, will just about wreck the whole world if you believe what some people say. Now whilst it’s true that in isolated instances and situations solar adds a new dimension of variability to the energy network here’s the simple fact – the grid is already a highly dynamic and variable beast.

Solar doesn’t wreck things, it just changes things.

A new study released today by the Australian Renewable Energy Agency looked at this very issue. Conducted by the highly esteemed CAT Project’s, the study used the Alice Springs grid network as a test case and in short asked two fundamental questions. The first was “What solar network penetration rates can be accepted without adversely and materially affecting network performance?” and the second was “What affect does geographic dispersion of solar have on the first question?”.

Now you may say “Why choose Alice Springs, one of the remotest towns on earth in the middle of a desert with a small population ?” Turns out that the weather, demand patterns and other factors make Alice Springs highly representative of a huge variety of Australian locations. It’s summers are hot, it’s winter nights very cold and incredulously it even rains there too.  It’s grid network is isolated, but as a simulation model of how a much larger National Electricity Market might behave it’s an ideal test location too.

The lessons learned

The first, important, mythbusting fact is this – network demand (ie load) is actually already highly variable. Network demand is not the stable, consistent, smooth profile some suggest but rather is made up of countless, constantly varying individual changes. This is pretty logical when you think about it because although their are load patterns, humans are a diverse lot who do things at different times for sometimes bizarre reasons and constantly changing random factors affect the resulting load pattern. Last night for example, our lights, washing machine and hot water all went on for an hour or so at around 2am (highly unusual)  while we sorted out poor old number three son’s upset tummy. The network saw a tiny little unpredicted change in load. Shit happens, as they say.

So, the study measured, analyzed and found in highly granular detail that load varies. So how does solar affect this?

One of the main reasons that Alice Springs was chosen for this project was because it was relatively easy to install a number of monitoring stations at a variety of distances from the CBD without intruding on too many people. Pretty much just a case of throwing your monitoring station over your shoulder and heading out of town then stopping at regular intervals to plonk them on the ground.

What this enabled the team to do was to measure and analyse solar insolation across a diverse geographical range. If clouds passed at one station (reducing solar generation), how long would it take to reach the next one under a wide range of conditions?

The findings come as no surprise to people who understand solar and who have been arguing its merits for eon’s. In some other countries where solar capacity is far larger, the measurement and aggregation of distributed wind and solar generation, load and weather data have been available for years. What these data points have shown is that natural geographic diversity has a smoothing effect; the sun may or may not shine in specific allocations, but when viewed as a whole there is a natural, organic tendency to balance each other out.

And that’s pretty much exactly what they found in Alice Springs.

In the case of Alice Springs, the acceptable spatial geographic diversity was around 5km. What the study found was that average cloud speed and size meant that at this distance, sufficient variation existed such that when one site was underperforming, another would typically be generating just fine. Chatting with the team at CAT Projects, they highlighted a couple of things from these findings. The first was that nature of networks varies and hence, the impact will also. Logically however, the larger and more diverse the network is, the less noticeable the effects of penetration will be, notwithstanding that some areas will be more stressed than others.  The second was that of course, weather patterns and cloud types vary around Australia.  What this means in reality is that instead of a cloud event taking 6 minutes to pass (for example), on the coast it might take 10 minutes, so the geographic range might change to 10km instead of 6km for example.

The results did show that (logically) if you plonk a huge 10MW wad of solar in one spot in Alice Springs the impact is worse; and it adds more variation to network conditions. This is not ideal although it can be mitigated through power control equipment and network design.

However, if you instead distribute that 10MW wad of solar over a wider geographic area in 1MW wads, the impacts are substantially mitigated – to the point where it is barely discernable over the natural organic variation in load that the network already see’s. In fact the results of the study found that it was “no worse than what the power system already accommodates”.

Scalability

Importantly, the results of this study do lend them self to scalability. The generation capacity, load and PV system size can in a relatively straightforward manner be scaled up to match the demand of far larger network areas; like the entire National Electricity Market for example. As both load and generation size and diversity increases, the impact’s become less intense – as long as the spatial rules are followed.

This highlights an issue that I’ve been arguing for many years (along with many others of course). That is, as soon as the networks get clever and start incentivising and planning solar uptake to match their situation, the sooner we’ll be able to provide even more benefits. A NEM wide solar rollout map with open access to network hotspots is the solution here but the solar industry can’t solve that one – its in the hands of the networks who right now, have little or no incentive to provide the data.

Getting this right is also a key part of the energy revolution in which generation diversity is planned, encouraged and systematically rolled out. It’s a matter of choreographing the technologies, variables and constraints and the intelligence is out there, as we highlighted in the case of Hawaii recently. If you want a visualisation of what this choreography looks like check this awesome video from one of the great TED talks.

Distributed energy 

This analysis doesn’t stop clean, centralised generation from making a whole lot of sense; it simply has to be planned well, timed perfectly and put in the right locations. The economies of scale are better and large projects can and should be deployed in the right conditions. However, as we have seen in the case of China for example, installing large volumes of solar in distributed sites is increasingly popular and in many cases far quicker. Recent data from China’s planners reveals that the 15GW solar target in their twelfth 5 year plan is now set at 8GW of central generation and 7GW of distributed generation.

What this ARENA study proved in an entirely statistically valid, Government supported study is this:

  • Distributed energy makes a whole lot of sense
  • Sufficient geographic spatial diversity overcomes many of the (previously assumed) problems
  • Network demand has a lot of natural variability
  • Well planned networks can cope with far higher solar penetration than is often assumed

The data from this study is publicly available for all network planners and I just bet the clever folks at CAT Projects would love to help develop a NEM wide roll out map.  Hat’s off to ARENA for providing funding support, CAT Projects and everyone else who participated in this awesome piece of work.

Content Discussion

Keith Pickering's picture
Keith Pickering on April 11, 2015

The real problem with high penetration of solar (or wind) isn’t variability, it’s cost.

1. When solar penetration exceeds the curtailment point — that point at which solar-plus-backup generation exceeds demand — generation must be curtailed or dumped to a neighboring grid, or overvoltage of the system can occur. This is actually a far worse situation than a blackout because it can cause destruction of grid-connection devices, fires, or even explosions. But when solar generation is curtailed, the capacity factor of solar takes a hit — and as a result the price of solar generation takes a hike.

2. Even when solar is not curtailed, high solar penetration in a deregulated market will cause a creeping economic death spiral. Since solar is time and weather dependent, when one solar panel is generating its maximum, chances are other panels in the region will be too. This simultaneous generation drives down the price of deregulated electricity, without affecting its cost. Thus the owners of solar generation will find themselves selling into a buyer’s market nearly all the time, getting rock-bottom prices for the generated electricity, while typically paying premium (compared to fossil) capital costs for the panels. This death spiral is unnoticable at low penetrations, but as more and more solar comes onto the grid, competing with more and more solar already on the grid, each incremental addition drives down received prices for existing solar owners — while their loans still need to be paid off at fixed rates.

For this reason, solar will not be economically viable in high penetrations unless protected by a regulated market. The problem there, as seen in Germany, is that there is a price to be paid for that regulation in the form of higher electricity prices and/or higher taxes. Somebody has to make those losses good.

Bruce McFarling's picture
Bruce McFarling on April 11, 2015

1. When solar penetration exceeds the curtailment point — that point at which solar-plus-backup generation exceeds demand — generation must be curtailed or dumped to a neighboring grid, or overvoltage of the system can occur. This is actually a far worse situation than a blackout because it can cause destruction of grid-connection devices, fires, or even explosions. But when solar generation is curtailed, the capacity factor of solar takes a hit — and as a result the price of solar generation takes a hike.”

This is why a country such as Australia, which has such a number of high quality solar resource regions, relative to its population, research and development continues on solar thermal CSP, where the solar power is dispatchable for a number of hours after collection.

Of course, the issue of solar curtailment is exacerbated by under-investment in transmission … the US’s failure to establish instititions to meet transmission needs on a regional and national scale is something we are going to have to tackle no matter what strategy we use to decarbonize our total energy supply.

2. Even when solar is not curtailed, high solar penetration in a deregulated market will cause a creeping economic death spiral. Since solar is time and weather dependent, when one solar panel is generating its maximum, chances are other panels in the region will be too. This simultaneous generation drives down the price of deregulated electricity, without affecting its cost.”

Also note that “deregulated electricity” is a misnomer … establishing artificial markets and mandating that they be used to set prices is not deregulation of electricity prices, it is simply a change in the way that electricity pricing is regulated. The full economic cost of solar power at various levels of penetration are an important factor in determining what penetration level is efficient … but whether or not “deregulated energy” pricing rules interfere with solar energy roll-out to an economically efficient level is not a factor for how much solar energy capacity we should roll out. It is rather a factor for deciding what pricing system we should use for that energy harvesting capacity.

Nathan Wilson's picture
Nathan Wilson on April 12, 2015

For all of the ideological appeal of residential scale PV “distributed generation”, the Australian (ARENA) report clearly shows prudently deployed centralized solar generation can work just fine (and we know centralized is always cheaper to build and maintain).  Beyond “The Simpson’s” TV show, no one builds grids with a single centralized generator, it’s always multiple generators paralleled for redundancy.

In fact, the report wasn’t about residential solar, but mearly pointed out that dividing a centralized solar facility into 9 distributed chunks would be a more effective strategy, unsurprisingly.  A similar study done for the European grid also pointed out that putting the solar generation on the high voltage grid (which is always done for utility scale projects) is even better, as it effectively averages it over the very large area served by the high voltage part of the grid.  In fact, much of the “distributed solar”  being installed in China is composed of MWatt-scale factory roof systems (with high voltage grid connections, and near-utility-scale economies of scale).

As Bruce points out in his comment,  adding energy storage makes solar even more useful by extending operation through the whole evening demand peak.  This also allows solar to actually replace fossil fuel power plants (rather than just reducing their fuel consumption), but only until the penetration reaches the level by which sunny-day electric demand exceeds cloudy-day demand (10-20% maybe?).

Storage also effectively removes the need to break-up centralized generation.  For an electrically isolated town with a grid demand equal to that of Alice Springs (53 MW peak); four centralized PV arrays with battery storage (and fossil fuel backup) would works just as well as dozens or thousands of distributed sites.  The batteries effectively deal with fast changes in PV output, so the fossil backup can keep up.

But grid studies of the effect of rising wind penetration in the US have shown that (at least at the GWatt scale) it’s always better to build-out long distance transmission before considering energy storage.  This won’t solve the problems of high renewable penetration at the residential level, but that is yet another reason utility scale generation is technically preferrable to residential.

But once we have transmission to towns like Alice Springs, why not combine solar with nuclear, and get rid of even more fossil fuel burning power plants?  Figure 6 in the ARENA report clearly shows that even in the studied part of sunny Australia, most of the annual electricity demand is baseload, and suggest an equal capacity of nuclear and solar would be near optimal.

Engineer- Poet's picture
Engineer- Poet on May 24, 2015

the issue of solar curtailment is exacerbated by under-investment in transmission

This “under-investment in transmission”… that would be “externalized costs of solar” under a deceptive heading, right?

the US’s failure to establish instititions to meet transmission needs on a regional and national scale is something we are going to have to tackle no matter what strategy we use to decarbonize our total energy supply.

The USA’s institutions and assets do a pretty good job of moving power from the existing plants to users.  What you mean is that those assets are neither designed nor suitable for transmitting power from far-flung “renewable” sources like Texan and Dakotan wind farms to major cities and industries, nor for time-shifting noon-peaking PV generation to evening-peaking demand.  Since nobody asked for that when the major systems were built out decades ago, it’s no surprise they do it poorly or not at all.

What is left out of your narrative is that the intermittent nature of the “renewables” isn’t suitable for the needs of most users, so almost every watt requires backup by dispatchable generators.  This makes the net cost far greater than what appears on its balance sheet.  If we are going to de-carbonize our power, we’re better off with nuclear power, which we can locate where transmission resources already exist and does generate on demand, not at the whims of weather and season.

http://bravenewclimate.com/2015/04/14/is-renewable-energy-looking-like-a-new-religion/#comment-401314

How much land has the almost 5GW of solar installed in Australia consumed? Essentially none. It’s all on rooftops.

So, what is its capacity factor?  What is its correlation of availability with demand?  In other words, what’s the use?

Willem Post's picture
Willem Post on April 19, 2015

Nigel,

This may be of interest.

The below site mentioned record PV solar output, MW, in Germany of 24,200 MW on 6 June 2014, and 25,100 MW on 15 April 2015, from 13:30 to 13:45, i.e., about 45 minutes!

http://www.germanenergyblog.de/?p=18424

PV capacity installed was 38,232.634 MW at end 2014.

System factors:

It is usually claimed individual PV systems have output of about 80% of installed capacity, due to various SYSTEM losses. Hence the maximum output could be about 38,233 x 0.80 = 30,586 MW.

Non-system factors:

Not all systems face solar south, are properly angled, are clean, are new, are unshaded. The sky may be partially clouded, or light may be obscured by pollution, snow and ice. All these factors, reduced the solar output, MW, by another (1 – 25,100/30,586) x 100 = 18% in April.

Alternatively, on an energy production basis, MWh/yr, the above system and non-system factors lead to about 14.5% less production/yr:

In 2014, Germany’s PV solar CF was 34,930 GWh/(38,236 MW x 8,760 hr/yr) = 0.104 out of a theoretical maximum in South Germany of 0.120, about 14.5% less.

Germany’s actual yield was 34,930,000 MWh/38,236 MW = 914 kWh/kW of panels. In southern Spain yields are up to 1,500 kWh/kW of panels.

The conversion ratio of PV panels was about 12% in 2002 and about 15% in 2014. 

http://www.ise.fraunhofer.de/en/publications/veroeffentlichungen-pdf-dateien-en/studien-und-konzeptpapiere/recent-facts-about-photovoltaics-in-germany.pdf

http://en.wikipedia.org/wiki/Solar_power_in_Germany