Earlier this week, Lawrence Berkeley Labs released a marvelous comparison of residential PV costs in Germany and the United States, finally putting some detail to an enormous gulf in costs (nearly $3.00 per Watt). The following chart (from page 35 of the presentation) shows the cost difference broken down into 9 categories, with ILSR’s addition… Continue reading
Viewing the economies of scale tag archive
In this April 2 presentation to the Pedernales Electric Cooperative of Johnson City, TX, ILSR Senior Researcher John Farrell discussed how solving the clean local energy puzzle requires much more than a consideration of cost per kilowatt-hour. Instead, cooperatives, municipal utilities and others considering developing local clean power should consider issues of scale, value and… Continue reading
Installed costs for solar PV have dropped and economies of scale improved significantly in 2010, opening the door for much more cost-competitive distributed solar power.
The data comes from the 4th edition of the excellent report from the Lawrence Berkeley Labs’, Tracking the Sun (pdf) and shows the installed costs for behind-the-meter solar PV projects in 2010. The following merely copies Figure 11 from that report, showing the average installed cost of “behind-the-meter” solar projects in the U.S. in 2010, by project size.
This is useful and shows the significant economies of scale for solar PV in 2010, but the history is important. For context, the following chart shows the 2010 data along with the 2009 data from Lawrence Berkeley Labs, with the grey shaded area indicating the cost decreases. The 2010 installed cost data from the California Solar Initiative (red) is also shown, helping validate the LBNL data. The last data point from the CSI is an outlier likely due to having too few projects in that dataset.
Two things are clear from the new data. First, installed costs have dropped significantly, by $1 per Watt for residential-scale solar PV and by nearly $2 per Watt for megawatt-scale projects. We can also see more clearly how the economies of scale of solar have improved, as well.
The unit cost savings between the smallest and largest solar projects (1 MW and under) jumped from $2.80 to $4.60 per Watt, a change in relative savings from 30 percent to 47 percent. Economies of scale were also much greater for mid-size solar (30-100 kW), with the percentage savings over the smallest projects rising from 21 to 35 percent. The following chart illustrates the change in economies of scale, showing installed costs as a percentage of the cost of a 2 kW system.
Instead of having relatively little economies of scale for solar PV projects larger than 2 kW, the 2010 data confirms that the unit cost of solar does continue to fall significantly as solar projects grow up to 1 megwatt (MW) in size.
Unfortunately, LBNL did not have sufficient data to provide context for economies of scale for larger distributed solar projects (1 to 20 MW), with only about 20 datapoints. However, their finding was that these larger crystalline solar projects cost between $4 and $5 per Watt, showing small but significant scale economies.
The lesson is that solar economies of scale seem to be improving as the U.S. market matures, good news for distributed solar to compete with peak electricity prices on the grid.
[note: for more context, see the previous post on 2009 solar economies of scale]
In August 2011, ILSR Senior Researcher John Farrell gave this presentation to a group of rural utilities and environmental organizations in Kentucky. The slides illustrate the enormous renewable energy potential in Kentucky and the cost-effectiveness of clean, local power in meeting the state’s electricity and economic needs. Clean Local Power for Kentucky View more presentations… Continue reading
Update October 2012: The 2011 Wind Technologies Market Report shows weak, but consistent economies of scale in wind power projects. It seems obvious: every extra turbine in a wind farm comes at a lower incremental cost, making the biggest wind power projects the most cost effective per kilowatt of capacity. If you bet $20 on… Continue reading
Update 7/26: One commenter asserts that the loss figures offered by the original author may be relevant in India, but do not reflect the U.S. grid, where losses total around 7%. EIA data seems to reflect this [xls].
Can transmission losses completely offset economies of scale for solar power plants? An article in Renewable Energy World argues against the building of multi-megawatt (MW) solar PV instead of on-site or local PV systems. In particular, the author writes:
The biggest problem with the multi-MW solar PV plant is that it loses 12-15 percent of expensive power as it passes through a series of power transformers. PV solar inverters generate power at 400 [Volts] three-phase. In large plants, this power is first boosted to 66 [kilovolts] or more with several power transformers and then stepped down to 400V with another string of transformers to suit consumer requirements. In addition, there is a further transmission loss of 5-7 percent in the power grid. Why suffer an avoidable 20 percent loss of expensive solar power?
…There is thus no ‘scale advantage’ in large PV solar plants. In reality, all multi-MW plants are basically clusters of several 500-kW plants since solar inverter capacities are limited to about 500 kW and no more. Why not have one hundred 500 kW plants instead of one giant 50 MW plant?
With 20% of the power from a large-scale solar plant lost in transformers and power lines, it could seriously alter our previous analysis of solar economies of scale. Here are the original charts, with the first chart shows our original analysis of solar economies of scale, with strong savings for scale for new projects (as reported by the Clean Coalition):
The next chart shows the economies of scale in the German rooftop PV market, as reflected in their feed-in tariff rates. The percentages show the price in each size tranche relative to the price for the smallest rooftop PV systems. Once again, there are significant savings for scale, especially when going from a project 100-1000 kW to one that is 1 megawatt or larger (15 percent).
But if there is a 20% power loss for the voltage stepping and transmission for larger solar projects, then when it comes to delivered power, small projects may perform better. Let’s assume that projects 1 MW and larger require the voltage step and transmission (and incur the losses), whereas smaller plants do not. The following two charts illustrate the difference.
The first chart takes the Clean Coalition (green line) data from the Solar PV Economies of Scale chart and calculates the levelized cost of the power from each size power plant based on the sunshine in southern California. For the largest size solar power plants, the cost is adjusted for the losses due to transmission and transformer stepping.
As we can see in the first chart, the losses from transmission wipe out most economies of scale for large-scale solar, making 1 MW and larger solar PV plants equivalent to on-site solar power from a 25 kW solar PV array.
We can similarly examine the effect in the German case. Here the government sets the price paid for solar by size class, and since it’s based on output at the power plant, large-scale plants that have transmission losses get paid for their entire power output, regardless of how much usable power reaches customers. The following chart shows what German customers effectively pay for solar, assuming that 1 MW and larger facilities all experience the 20% transmission losses explained earlier.
As we can see in the chart, the cost of transmission can wipe out the economies of scale in installed costs, making large-scale solar comparable to solar PV of 30-100 kW, but without the same transformer and transmission losses.
It may be true that the installed costs of solar PV continue to fall as projects get larger, but it’s clear that relying on the price of solar at the power plant does not accurately reflect the cost to the grid or ratepayers. For some size of larger power plants (1 MW? 5 MW?), the lost power from stepping up and down voltage through transformers and from transmission may largely offset the economies of scale from building a larger power plant.
Rather, mid-sized solar (or specifically, projects that can connect directly into the distribution system without changing the voltage) may deliver the best cost per kilowatt-hour.
CSP represents over 6,000 MW of the over 15,000 MW of future solar projects that SEPA is tracking, but there are differences in project development between CSP and PV. PV can be built and sub-sections of the larger project can be energized over time, resulting in lower construction risk and balance-sheet impact. CSP projects need to be completed in full before commissioning, a period which takes several years from start to finish. [emphasis added]
PV projects, which ranged in size from 1-kilowatt residential installations to 48-megawatt power plants, have much shorter planning horizons and project completion times, along with lesser siting, permitting, financing and transmission requirements at these small- and medium-sized scales.
However, larger PV and CSP projects (those greater than 50 MW) require overcoming financing, siting/permitting, and transmission barriers that might emerge at these larger sizes. [emphasis added]
In drafts of ILSR’s forthcoming report on a distributed generation future (check back June 22!), I took some flak for my solar PV economies of scale analysis. In it, I used data from the California Solar Initiative (through 2009) to point out that most economies of scale in solar PV seem to be captured at a size of 10 kilowatts (a large residential-scale project).
“The solar statements seem way off base,” wrote one reviewer.
Upon further review, I stand by my initial claim. But, I note that the critics have a point, as well.
For deeper analysis, I grabbed data from Lawrence Berkeley Labs’ 2010 report Tracking the Sun III, which provided a very nice breakdown of installed costs for solar PV by project size. I then dropped those size ranges into the California Solar Initiative (CSI) data for the whole data set (2006-2011) as well as for just the past two years (2010 to present). The following chart illustrates the findings:
The historic data confirms my earlier analysis, that most economies of scale are achieved at small size. In the full CSI database, there’s a 23% decrease in per Watt cost when increasing project size from under 2 kW to 5-10 kW, but only a further 6% percentage point decrease in sizing up to over 1,000 kW. The other two curves are quite similar.
But the historic U.S. data is not the only story.
The Clean Coalition – a distributed generation advocacy organization – has different numbers on installed cost from their network of installer partners. These figures, data on very recent or proposed installations, tell a different tale, illustrated below:
In the Clean Coalition data, the savings from 5 kW to 25 kW are about 10%, but the savings from upsizing to 100 kW are a cumulative 21%, and growing to 1,000 kW offers a total of 28% off the 5 kW price per Watt. In other words, economies of scale continue strongly through the 100 kW size range.
Their data is not alone. In the German feed-in tariff, solar PV producers are paid a fixed price per kWh generated, with prices set according to the location of the solar PV plant (roof/ground) and by size (small, medium, large, etc). Overall, Germany is simply cheaper, with average installed costs for 10-100 kW rooftop PV installations of just $3.70 per Watt. But their economies of scale are also strong: there is a 10% price differential between rooftop solar arrays smaller than 30 kW and those 100-1000 kW, but an additional 15% price drop for projects over 1000 kW.
The conclusion is murky. Historical data in the U.S. supports my original assertion: economies of scale for solar PV are limited beyond 10 kW. But recent installed cost data and the German experience both suggest that there are stronger economies of scale up to projects 1,000 kW (1 MW) in size.
While seeming counterintuitive, a focus on smaller-scale distributed generation enables more and faster development of cost-effective renewable energy.
Last week I wrote about the illusion that we can “move forward on all fronts” in renewable energy development; rather, a bias toward centralized electricity generation in U.S. policy reduces the potential and resources for distributed generation.
In contrast, distributed generation provides unique value to the grid and society, and its development can also smooth the path for more centralized renewable energy generation.
First, distributed generation is cost-effective. Economies of scale for the two fastest-growing renewable energy technologies (wind and solar) level off well within the definition of distributed generation (under 80 megawatts and connected to the distribution grid). Solar PV economies of scale are mostly captured at 10 kilowatts, as shown in this chart of tens of thousands of solar PV projects in California. Wind projects in the U.S. are most economical at 5-20 megawatts, illustrated in a chart taken from the 2009 Wind Technologies Market Report.
Besides providing economical power relative to large-scale renewable energy projects, distributed renewable energy generation also has unique value to the electric grid. Distributed solar PV provides an average of 22 cents per kWh of value in addition to the electricity produced because of various benefits to the grid and society. The adjacent chart illustrates with data coming from this analysis of the New York electric grid. Grid benefits include peak load shaving, reduce transmission losses, and deferred infrastructure upgrades as well as providing a hedge against volatile fossil fuel prices. Social benefits include prevented blackouts, reduced pollution, and job creation.
Distributed wind and solar also largely eliminate the largest issue of renewable power generation – variability. Variability of solar power is significantly reduced by dispersing solar power plants. Variability of wind is similarly reduced when wind farms are dispersed over larger geographic areas.
Not only are integration costs reduced, but periods of zero to low production are virtually eliminated by dispersing wind and solar projects over a wide area.
As mentioned at the start, distributed generation also scales rapidly to meet aggressive renewable energy targets. Despite the conventional wisdom that getting big numbers requires big project sizes, the countries with the largest renewable energy capacities have achieved by building distributed generation, not centralized generation. Germany, for example, has over 16,000 megawatts of solar PV, over 80 percent installed on rooftops. Its wind power has also scaled up in small blocks, with over half of Germany’s 27,000 megawatts built in 20 megawatt or smaller wind projects. In Denmark, wind provides 15-20 percent of the country’s electricity, and 80 percent of wind projects are owned by local cooperatives.
With all these benefits, distributed generation can also smooth the way for centralized renewable energy, in spite of energy policies that favor centralized power. When distributed generation reduces grid stress and transmission losses by provided power and voltage response near load, it can defer upgrades to existing infrastructure and open up capacity on existing transmission lines for new centralized renewable energy projects. A focus on distributed generation means more opportunity for all types of renewable energy development.
It may seem counterintuitive, but distributed renewable energy should be the priority for reaching clean energy goals in the United States.