Giant, remote desert solar installations are often touted as the solution to our fossil fuel addiction, especially as a way to replace coal-powered electricity generation. It is true that in the course of operation, industrial solar power generating stations cause far less direct greenhouse gas (GHG) emission per kilowatt-hour of generated power than do coal-burning and other fossil-fueled plants. However, industrial solar plants are by no means GHG-neutral. In the course of their construction and normal operation these plants have their own significant greenhouse gas footprint, which must be accounted for before we can really gauge the prospective climate benefit of any such project. Before a project can provide a benefit to the planet’s climate, in other words, it must first amortize the greenhouse debt it has incurred simply in its construction and operation. To date, very little discussion of this greenhouse debt has taken place, leading to a distorted impression of industrial solar plants’ climate benefit. This briefing paper is intended to help correct those distortions.
Capacity Factors and Transmission Losses
Before we describe how industrial solar plants incur their greenhouse gas debts, we should clarify common misunderstanding about the rate at which those plants pay off their debts with “carbon-free” energy. Industrial desert solar plants are routinely described as having projected output many times higher than the likely actual figure. This is because many descriptions of the plants’ capacity fail to take into account their “capacity factors,” which is the factor by which one must multiply their rated maximum output to take into account times when the plant will be unable to generate power at the optimal capacity.
For example, the proposed Ivanpah Solar Energy Generating System is billed as a 370 megawatt (MW) generating plant once it is completed. That figure is accurate, in that the plant probably will generate 370 megawatts of power around noon on a cloudless day in mid-summer, when the sun is as close to directly overhead as possible. But when the sunshine striking the plant is less energetic, for instance during winter, on cloudy days, during night, early morning and late afternoon hours, the plant’s output will be significantly less. During high winds, which happen frequently in the Ivanpah Valley, the mirrors will need to be secured, rendering the plant inoperative. The plant will likely need frequent maintenance, which reduces output during times when it is offline. Utility experts refer to this kind of correction for actual real-world output as a plant’s capacity factor. Coal and nuclear plants commonly have capacity factors in the 60-90% range: they can run non-stop near peak output for weeks on end. Solar plants’ capacity factors run closer to about 25-30% at best. The California Energy Commission estimates the likely capacity factor of the Ivanpah SEGS as around 28%.
For this very reason the figures often cited in news articles referring to solar plants’ output with a phrase such as “enough to supply 400,000 homes” are almost always misleading, if not completely incorrect. To determine how much generating capacity is needed to power a certain number of homes, the relevant unit isn’t megawatts, but megawatt-hours. A 100 MW solar plant with a capacity factor of 30% will produce only a third the megawatt-hours produced by a 100 MW coal plant with a 90% capacity factor. Most homes don’t use electricity only when the sun shines; in fact, they often use more when it doesn’t. Tying solar-plant power generation to a certain number of homes is largely meaningless. (Conversely, since rooftop PV systems are often used in conjunction with batteries to store energy, you can say each rooftop produces enough electricity to power a home).
Adding to — or more accurately, subtracting from — the issue of lower-than-anticipated delivery of energy from these sites is the problem of transmission inefficiency. As high-voltage electricity flows through transmission lines, a significant percentage of its power is lost due to resistance in the conductor. Average transmission losses in the North American grids are somewhere around 6.5-7%. Moreover, resistance in a transmission increases as its temperature rises, and summer temperatures in the deserts are routinely 20-30 degrees F higher than elsewhere on the continent, with afternoon highs above 115F not uncommon. Under such circumstances, losses in power transmitted from desert industrial solar plants may be considerably higher than the US average.
The concrete used in solar installations is a large contributor to the GHG burden. Concrete is a mixture of sand, gravel or aggregate, and cement, which is produced by heating limestone to about 1450°F. This process contributes a large amount of CO2 to the atmosphere, both from the energy consumed to heat the limestone and as the heated limestone gives off CO2. The cement industry is the second largest CO2-emitting industry, after power generation, producing about 5% of global man-made CO2 emissions. The amount of CO2 emitted by the cement industry is nearly 900 kg of CO2 for every 1000 kg of cement produced. High-quality concrete used in industrial applications may be as much as 25% cement. Such concrete weighs approximately 2400 kg per cubic meter, and thus each cubic meter of concrete used in construction of solar power facilities contributes 540 kilograms of CO2 to the atmosphere solely from the production of the cement included. That’s more than the maximum amount of CO2 the state of California allows a coal-fired plant to emit per each megawatt-hour of energy produced.
A cubic meter of concrete is a cube measuring a meter (about 39 inches) on a side. A standard cement truck holds six cubic meters. Concrete will be used throughout most remote solar sites in significant quantities as footings for equipment, foundations for buildings, berms and culverts to divert flash floods, and for other purposes.
In addition, transporting the concrete from supply yards to the construction site incurs a significant additional CO2 contribution, especially at remote sites not served by rail.
It is not only concrete that must be shipped to the site of a large desert solar installation, but all building materials not extracted from the site itself must be shipped in, often across considerable distances. Some sites of proposed public-land renewable energy projects in the desert Southwest are 50, even 100 miles from the nearest communities with substantial workforces. This means that unless barracks housing is built on the site — with additional environmental cost — commuting workers add to the CO2 debt the project must be accounted for, especially during the construction phase.
One of the least-known greenhouse consequences of industrial-scale solar generating facilities — or indeed of any high-voltage industrial facility — is the use of sulfur hexafluoride, or SF6. SF6 is an inert, non-toxic gas five times heavier than air that is used as a reliable gaseous insulator in high-voltage transformers, circuit breakers and switchgear.
Though it is inert and thus non-toxic, SF6 does pose a toxicity risk to people and wildlife under certain circumstances. If exposed to high-voltage discharges SF6 molecules can be bonded, forming disulfur decafluoride — a highly toxic gas once considered for possible use as a chemical weapon.
The main danger SF6 poses, however, is to the climate. SF6 is 23,900 times as potent a greenhouse gas as CO2, making it the most potent greenhouse gas the Intergovernmental Panel on Climate Change has evaluated. In other words, a pound of sulfur hexafluoride contributes as much to global warming as 11 tons of CO2. SF6‘s overall contribution to the global GHG problem is relatively small, likely contributing to less than one percent of total warming from all sources. But in the context of power generation projects which purport to reduce greenhouse gas emissions, SF6 emissions become particularly relevant — especially if those projects require long-distance transmission, which accounts for around four-fifths of SF6 emissions.
There are almost no natural “sinks” for SF6 — it is not taken up by living things or absorbed by water, soil or rock — so it stays in the atmosphere until it breaks down, which can take as long as 3,200 years.
SF6 leaks out of pressurized switchgear and circuit breakers over time, and is also released to the atmosphere during maintenance, replacement of equipment, and — most notably for determiningthe greenhouse gas burden of new industrial-scale energy projects — during installation of new equipment. While older circuit breakers may hold up to 2,000 pounds of SF6, more modern designs such as those that would be installed for new generation and transmission facilities cut that to around 100 pounds. The EPA estimates that even the most aggressive leak detection, repair and recycling programs could only cut SF6 emissions by about 30 %.
As most SF6 emissions are generated in long-distance transmission of electrical power, the more remote a new facility is, and the more additional miles of transmission line needed to deliver its power to the grid, the higher the SF6 burden of each new generating facility will be. In 2010 the EPA estimated average emissions of between .58 and .89 kilograms of SF6 for every mile of transmission line per year over the last decade.
As an example of direct impacts of industrial solar projects on SF6 emissions, consider the proposed Tessera Solar project at Calico, in California’s Mojave Desert, which would place tens of thousands of solar-heated Stirling engines on 8,230 acres of public land. Tessera itself projects that through routine leakage from circuit breakers on the site, its project would emit 36 pounds of SF6 into the atmosphere each year — the equivalent of 384 tons of CO2. The actual figure, given wear and erosion of seals in the desert’s intense heat, UV radiation, and wind-driven sand, is likely significantly higher—and that only accounts for emissions from the project site. In order to transmit the energy created at Calico to the grid, the adjacent Southern California Edison Pisgah Substation would be expanded from five acres to 40 acres, an eight-fold increase in size and likely similar increase in routine SF6 emissions.
Distributed generation, naturally, by drastically reducing the need for new transmission lines and substations and perhaps even allowing the decommissioning of older lines, offers an opportunity to curb SF6 emissions dramatically in the long run.
Water use at industrial solar facilities is mainly a concern due to oversubscribed streams and aquifers. The additional demand on local water supplies by big solar plants often puts wildlife in danger of losing the springs and seeps on which they rely. Yet there is a greenhouse gas aspect to water as well. In the arid and semiarid West, prodigious amounts of energy are used to move water from one place to another. In California, fully 20% of the state’s electrical consumption results from the pumping, transport, purification, and treatment of water. Energy used to transport water to industrial solar facilities will either be supplied from non-solar sources, in which case it adds to the facilities’ greenhouse debt, or it comes from the facilities themselves, in which case that power cannot be counted toward paying down that debt.
In the deserts, excepting those places adjacent to the already oversubscribed Colorado River or its aqueducts, water used in industrial solar facilities will almost always be pumped groundwater. (A preposterous example of this is Tessera’s proposed Calico plant, the water for which would be pumped from the Cadiz aquifer sixty-five miles away, and then shipped to the site by rail.) In 2005, Robert Wilkinson of the University of California, Santa Barbara estimated the typical energy consumption of wellwater pumping in southern California’s Chino Basin at 2,915 kilowatt-hours per million gallons, a figure likely similar to those for the deserts’ deeply buried aquifers.
Water consumption at industrial solar power plants varies widely by the type of plant. Concentrating solar thermal facilities may use water either to drive steam turbines or for cooling, or both. Conventional solar thermal technology is found in both “wet-cooled” and “dry-cooled” forms. The amount of water a wet-cooled plant uses may be considerable. The existing Solar One plant in Boulder City, Nevada, a wet-cooled 64-megawatt capacity solar trough field, consumes about 400 acre-feet each year. An acre-foot of water is that amount that would flood an acre of land a foot deep, or 325,851.4 gallons. To pump the 400 acre-feet of water Solar One uses each year, if we use Wilkinson’s Chino Basin pumping energy consumption estimate, would consume close to 380 megawatt-hours of power.
Dry-cooled plants use ambient air to cool their boilers rather than water. This saves a considerable amount of water, but not without cost. Dry-cooled facilities often involve more extensive infrastructure to channel convecting air, and they become less efficient as the air temperature rises — which is, ironically, not only the time when the facility would ideally be most productive, but also the time when electric power demand is likely to peak.
Even dry-cooled plants still use a lot of water. The Ivanpah Solar Electric Generating Station, which would be a dry-cooled plant, is expected by its promoters to consume at least 100 acre-feet of water each year —an eighth of the total annual water budget in the Ivanpah Valley.
Whether they are photovoltaic or concentrating thermal in design, industrial solar facilities depend on regular cleaning in order to run at peak efficiency. Even a thin layer of dust on PV panels or mirrors can cut output by a considerable amount. The aforementioned project at Calico, which would involve shipping water by rail, is billed as a “no-water-use” project using Stirling engines, but despite those engines’ putatively water-free operation the project’s 34,000 solar dish mirrors will still need cleaning, and this — along with fire protection, creation of hydrogen for the Stirling engines, and drinking water for workers — is why water would be shipped from Cadiz. Though research continues into dust-repellant coatings and dry cleaning methods, getting rid of dust requires hosing down the relevant pieces of equipment. The cleaning method may add a significant amount to the facility’s greenhouse debt. At Ivanpah, plans are to drive diesel-powered water trucks through the facility every so often to spray the mirrors, adding that diesel fuel to Ivanpah’s GHG debt.