I’ve heard arguments lately for local photovoltaic solar power (PV) from rooftops, roadways, and parking lots as a primary source of electric energy, mostly accompanied by arguments against long distance high-voltage transmission lines (HVDC). I keep picturing a revised Treasure of the Sierra Madre with bandits telling Humphrey Bogart: “Transmission lines? We don’t need no stinking transmission lines!”

I think the key to this argument is whether you are satisfied with slow incremental growth in renewable energy that gradually rises to providing 20 percent of electricity use, or if you want renewable electricity use to grow large enough to displace coal, natural gas for electricity, and even natural gas for heating and oil for transport (via ground source heat pumps and electrified transport).

Let’s look at data from the Carnegie Mellon Electricity Industry Center for one [PDF] PV system for one day in Prescott Arizona.


click image to zoom

You will note that in this example from one of the most suitable spots for solar electricity in the U.S. (a desert with both strong sun and few clouds), most of the power is delivered in an 8 hour period. In a city further north or with more typical cloud cover that would be more like five hours. Even so, that leaves 16 hours a day during which solar PV can supply little or no power. There are also short periods even during peak sun when cloud cover reduces solar power to near zero. PV without storage can only supply a small minority of electric demand. The least-expensive site-independent storage method, gigawatt-scale flow batteries, run about $350 per kWh. So seven hours of storage, about the minimum to significantly increase reliability would run $2,450 per KW. Sixteen hours storage, which would provide true (albeit not overly reliable) baseload, would cost $5,600 per KW. As long as electricity storage costs remain that high, PV is likely simply be used as supplementary power. If we don’t bring in complementary distant renewable energy, the most likely main power source will remain coal, natural gas, or nuclear energy. We are simply not likely to add that high a capital cost to already expensive solar cells.

And contrary to misleading information out there, solar PV is still expensive. The lowest price for PV systems where reliable cost information is available is $5,600 per KW of DC PV installed in Germany. (Note that part of that is electronics, installation, not just the solar cells themselves.) At a 5 percent interest rate and 21 percent capacity factor, that yields a $0.24/kWh price. (Note that the lowest price we know of in California is $7,100 per DC kWh installed, which would yield a $0.31/kWh price.) So where do lower price estimates come from? Many estimates include tax breaks, or don’t include interest, or assume maximum lifespans.

Sometimes the errors are more subtle. For example a lot of over-enthusiasm has been generated by estimates that the Sempra El Dorado PV project produces solar electricity at a cost competitive with coal. This estimate comes from an equity analyst who, like all of us outside of Sempra, does not have access to actual cost data. I will note that Michael Allman, Sempra’s Chief Executive says that the cost of electricity from the PV project is “more expensive than the power produced next door by burning natural gas.”

Now it is still likely that the Sempra El Dorado project is producing solar electricity at a lower price than your average PV plant. After all they have the connections to obtain the least expensive solar panels in the business, and with a ten megawatt order probably got the maximum quantity discount as well. Because this is a solar farm, the sort of project radical decentralists oppose, installation costs are minimized by building on ground level, tearing up existing plants and greenery, rather than incurring the high cost of installation on a rooftop, a road retaining wall, or on poles shading an existing parking lot or roadway. Maybe they obtained panels at as little as $3,000 per KW, and incurred total costs (including installation) as low as $4,000 per KW. That would yield a $0.17/kWh. This is an over-optimistic low estimate; the real number is almost certainly higher.

If you go back to that Carnegie study, you will also find that the December production for the largest PV installation is slightly over 54 percent of June output output. December demand for APS, the company who owns this site, is more than 78 percent of summer [XLS]. That still leaves slightly less than a one-third gap between winter output and demand. More capacity, rather than storage, is the answer to this kind of long term gap, so what capital cost is being increased by a third is an important issue.

Note that wind costs about one-fourth solar PV per kWh. Some wind farms, constructed in unusually favorable circumstances, ended up with total costs of four cents per kWh. The El Dorado project by the most optimistic reasonable estimate cost $0.17 per kWh. A typical big wind farm with favorable (but not exceptionally favorable) circumstances has $0.06 per kWh costs. A comparable solar PV project yields $0.24 per kWh costs. In less favorable circumstances, power from a wind farm cost $0.08 cents per kWh. Comparable solar PV costs $0.31 cents per kWh. Even if you ding wind 1.5 cents per kWh for transmission lines and transmission losses, you still end with solar PV costing more than 3.5-times what wind does. Now that does not mean there is not a place for solar electricity, and even for PV. If we are serious about phasing out fossil fuels, we can justify PV as a way to lower total system cost, and replace the maximum amount of fossil fuel. But we have to look at it as system, one that includes connection to other renewables.

For example, a recent study [PDF] by the New Rules Institute suggests that about half the states in the U.S. could supply all their own needs from renewables, and the rest could supply a substantial percentage. This is often offered as an argument against the need for long distance transmission, but actually supports it. A situation where some states can produce all the energy they need, and a bit besides, and others need more than they produce is actually the situation where you would need transmission between states. And it may be true that a state able to produce its own wind is better off doing so than importing electricity from where wind resources are better, due to transmission costs. But it is equally true that a state saves money by importing wind power compared to producing solar electricity. It is also true that this study relies a great deal on agricultural production of biofuel — which in a lot of cases does not produce net reductions in emissions, or produces very small reductions in emissions, and which competes with food production.

Incidentally, if we were looking only at per kWh cost, there would be a good argument to make for close to a 100 percent wind grid, which would need a lot of long distance transmission to get wind to low-wind states. Even with line, losses, transmission losses, and so forth, the highest wind costs would still be lower than cheapest solar costs. The reason for including solar is not to reduce transmission but to increase reliability and minimize the storage need to
make a renewable grid reliable.

Just as there is more solar energy available in summer than in winter, there is more wind energy available in winter than in summer. Just as demand in states that are optimum for solar production peaks in summer, demand in states that are optimum for wind production peaks in winter. But, as with solar, wind production drops more in the summer than electricity demand does. So connecting states with high wind capacity and states with high solar capacity reduces needs for capital buildouts in both states. It reduces the number of wind generators and the number of solar panels needed.

More importantly it reduces the need for storage to cover daily drops in production. Both wind and sun are variable sources. The graph above shows not only that huge gap from well before sunset to well after sunrise, but temporary drops during peak production periods. Similarly wind tends to speed up, slow down, start and stop. According to an RMI study [PDF], interconnecting wind energy with solar energy produces more reliable power than either alone. What none of the studies focus on, but is what is most certainly the case, is that interconnected wind and solar decrease storage needs by much more than they decrease variability. Imagine for the moment a power source that pauses for one continuous hour a day. Imagine a second that pause for a half hour, three times a day. The second source loses more power to pauses per day than the first, and pauses more often. But you can bridge that gap for the second with a half hour of storage, whereas it would take an hour of storage to bridge the gap for the first. Wind and solar complement one another significantly. They each produce most strongly when the other is weakest. The average length of low production periods should be reduced much more than the reduction of total low production time, allowing more of these gaps to be bridged by storage and less by backups.

But this effect will require transmission. The strongest solar resources and the strongest wind resources tend to lie at a distance from one another. And of course there are many states that will have to be net importers of renewable electricity.

Now even so, this would tend to imply much more wind than solar energy in an optimum grid. Even though solar resources are more plentiful, they are more expensive to tap per kWh. And to some extent that is true. A lot of energy is used to generate low-temperature heat for space and hot water heating, and for certain commercial and industrial purposes. We may be better off applying direct solar energy to these purposes than over using them for electrical generation.

But there are two current technologies that may tip use of solar energy to something comparable to wind. First, we can use concentrating solar power (CSP). This uses mirrors to focus solar energy the same way we can use a magnifying glass to set fire to a piece of paper. The heat from this can drive heat engines. Not only is this less expensive than PV, but we can also store heat for later use at a cost $35 per kWh, about one tenth the cost of storing electricity. And if that stored energy is used comparative quickly, within a day or so, net thermal losses are smaller than round-trip losses in electrical storage.

If we use solar energy for electricity but only for peaking purposes, relying mainly on dispersed wind mixed with a tiny fraction of that solar energy for baseload, then we don’t need thermal storage. In that case CSP can focus sunlight on advanced (40 percent) efficient solar cells designed for aerospace use. It is at least possible that the overall cost per kWh might be even less than driving heat engines, maybe even low enough that solar would be less expensive rather than more expensive than wind, even with today’s technology. You no longer have the storage advantage, but if it was cheap enough you could “overbuild” and end up with a mixed wind/sun grid with a lot of solar electricity — one that needed very little storage in return for producing some electricity that could neither be stored or used, but simply had to be discarded. (In practice, someone would find a use for “waste electricity” something low value, but worthwhile with electricity that was truly surplus.) I emphasize this is speculation. I don’t know that we could do it. But a very smart engineer who posts under the name of Sunflower on this board has argued for very low overall costs low from combining CSP with space solar cells. If he is right, then replacing storage by excess generation follows. (And you would still use substantial wind, and some storage, but it might shift the optimum solar/wind balance closer to 50/50.)

While I think there may be real conflict between this and what radical decentralists say, I also think this conflict can be exaggerated. This is not an argument that the majority of power has to travel long distances. I think a lot power can be generated locally, depending on what you define as locally. Only a minority of power will have to travel long distances, and only a minority of that will need to travel more than a few hundred miles.

Saying that we will need transmission to convert to a mostly renewable society is not the same as saying that most of the power lines proposed by current utilities are justified. Carol Overland, in her recent Gristmill post made the point that most (perhaps all) such proposals are about merchant power to buy and sell more conventional generation, mostly coal and nuclear power.

The Sunrise Powerlink proposal is a good example that substantiates this. Looking at the EIA we find that constructing the 134-mile one gigawatt line would emit about 110,000 tons of CO2 equivalent, and that SF6 leaking from the gas insulated lines would produce even more emissions over the lifetime of the line. According to the EIA it would take 12 years just to pay back the construction emissions. Now if the line was actually used to significantly increase net renewables, payback would not be so absurdly slow. For example, if 10 percent of the line’s capacity was used for net additional renewables, it would take about eight days to pay back construction CO2, and perhaps another 8 to 16 days to pay back all SF6 likely to be emitted over the life of the transmission lines. (This assumes each kWh displaced saves slightly less than 1.3 pounds of CO2 equivalent.)

The bottom line is, just as critics say, the Sunrise link is designed to bring additional coal power to San Diego, and secondarily to connect to the Stirling solar project which most critics judge likely to fail, thus giving PGE an excuse to not meet the its renewable portfolio requirement. An early sign that this was some sort of fraud was contracting solar power from Arizona, whose quality of solar resources differ trivially from those of San Diego and surrounding areas. Arizona does have substantially greater geothermal resources, which PGE is less eager to exploit.

One possible alternative offered to renewables is recycled energy, where industrial waste heat is used for electricity generation. This is a great resource that should be exploited. But it does not eliminate the need for a mostly renewable grid. For recycled energy to produce a high percent of our electricity would require industrial burning of fuels to stay close to our current level. According to Tom Casten, total recycled energy potential is about 20 percent of current electrical demand. If we reduce industrial fuel use substantially, we will also reduce the potential for recycled energy to much less than this. That does not make it an unimporta
nt resource or one we should not tap. There are a few cheaper means of saving energy, (cough*weather.sealing*cough), but not many. We should take full advantage of it to the maximum extent we can. But, if we seriously intend to reduce greenhouse gases we must reduce both industrial and electrical emissions. Recycled energy is a supplement to, not a substitute for, a grid that includes long distance transmission (including some very long distance transmission).