Saturday, February 7, 2015

Costs and Choices - Fermi 3

DTE Energy, formerly known as the Detroit Edison Electric Company, wants to build a new nuclear power plant, Fermi 3, next to Fermi 2 and the ruins of Fermi 1. The design of Fermi 3 is for approximately 1.55 billion watts of electrical energy output. It sounds like a lot of energy, and it is reasonable to wonder what this might cost.

DTE estimated in their 2008 application for an operating license for Fermi 3, that the construction cost would be $3,500 - 4,500 per kilowatt of electrical output, or $3.50 - 4.50 per watt. Applied to the design for 1.55 billion watts of electrical output, that would be a range of ~5.5 to 7 billion dollars. We can take this as a low estimate in 2008 dollars.

From the Wikipedia article "Economics of nuclear power plants":  "In Canada, cost overruns for the Darlington Nuclear Generating Station, largely due to delays and policy changes, are often cited by opponents of new reactors. Construction started in 1981 at an estimated cost of $7.4 Billion 1993-adjusted CAD, and finished in 1993 at a cost of $14.5 billion." In round figures, this would mean that taking DTE's original estimate and doubling it to $14 billion would be closer to a real figure.

From a Physicians for Social Responsibility 2008 report, "Nuclear Power Plant
Construction Costs":  "... total costs (including escalation and financing costs) will be in the range of $5,500/kW to $8,100/kW ..." Applied to the 1.55 billion watts for Fermi 3, a range of 8.5 to 12.5 billion dollars results. it is not clear if this includes the costs of financing construction, which would be paid after the plant starts operating.

A 2010 article from Greentech Media ("How Much Does Nuclear Cost? $6,000 a Kilowatt or More") discusses a cost of $6,000 per kilowatt and points to a study showing the "all-in" costs, including the costs of financing, would be in the $10-12.50 per watt range. That puts the high estimate for Fermi 3 over $19 billion.

The estimates above are from the period of 2008-2010. There's been a bit of inflation since then. There's also been a horrendous multi-reactor meltdown at Fukushima since then, which reasonably ought to result at least in additional safety requirements, which would raise the cost.

We have a range of estimates from a low of 5.5 billion 2008 dollars to over 19 billion dollars. Doubling DTE's upper figure of $7 billion 2008 dollars and allowing a bit for inflation since 2008 would result in a rough estimate of $15 billion. $15 billion, which this article uses, is a very approximate but reasonable figure to work with for the purpose of comparing building Fermi 3 to an alternative use for the money.

This is public money we are talking about, not (at least not yet) DTE's money. DTE expects to pay for construction through a combination of federally guaranteed loans and electrical rate hikes approved by the Michigan Public Service Commission. Whether it comes from federal taxes or rate hikes, that's our money, and we should decide how it will be spent.

The installed cost of solar panels for large-scale projects is now around $3.00 per rated watt. This is again a rough figure, with variations according to the scale of the project, local permitting costs, labor costs, etc., etc.

With solar panels, we can expect electrical power output equivalent to 4-6 hours of rated wattage per day in most US locations. That would be compared to an average 21-22 hours output per day for a nuclear reactor. In other words, the average daily power output from a nuclear reactor rated for 1 megawatt would be 4-6 times the power output of a solar farm rated for 1 megawatt, depending on the location of the solar farm.

I've worked up a table showing one possibility for spending $15 billion over 12 years on large-scale solar panel installations:



installed installed B watt-hrs B watt-hrs B-watt-hrs
year $ (B) $/watt B watts per day (new) per day (cum) per yr, (cum)

solar solar solar solar solar solar







1 1 3.00 0.333 1.410 1.410 514.650
2 1 2.95 0.339 1.434 2.844 1038.023
3 1.1 2.90 0.379 1.604 4.448 1623.659
4 1.1 2.85 0.386 1.633 6.081 2219.570
5 1.2 2.80 0.429 1.813 7.894 2881.262
6 1.2 2.75 0.436 1.846 9.740 3554.986
7 1.3 2.70 0.481 2.037 11.776 4298.369
8 1.3 2.65 0.491 2.075 13.851 5055.779
9 1.4 2.60 0.538 2.278 16.129 5887.137
10 1.4 2.55 0.549 2.322 18.451 6734.795
11 1.5 2.50 0.600 2.538 20.989 7661.165
12 1.5 2.45 0.612 2.590 23.579 8606.441








15.000
5.574
23.579 8606.441

B$ total
B watts
B watt-hrs B watt-hrs



total
per day total per yr total








15.000
1.55 Equivalent for nuclear
12220.200

B$ total
B watts

B watt-hrs



total

per yr total







. “B” is used to mean “Billion.”

“(cum)” is used to mean “cumulative.”


I've assumed a steady but not spectacular drop in installed solar panel costs per watt from $3.00 to $2.45 over the period. This is a conservative estimate, in contrast to a "target" of $1.00 per watt (Google SunShot Initiative) which might prove to be unrealistic. I've also used the 4.23 hours per day annual average output factor provided by the national Renewable Energy Lab's PV Watts calculator (http://pvwatts.nrel.gov/pvwatts.php) for the Detroit area.

The result? $15 billion could give us just over 5.5 billion watts (rated) of solar electricity. In terms of power, with the conservative assumptions used, the annual average watt-hours from the new solar installations would be a bit over 2/3 of the average annual watt-hours from Fermi 3.

Now, if the assumptions were changed in favor of faster improvements in solar panel efficiency and a more rapid lowering of installation costs, the expected annual power output would increase. 5.5 billion watts of peak power is pretty near the minimum that should be expected from a program to spend $15 billion installing solar panels in Michigan. But even with this minimal expectation, the solar project is well worth doing.

It's important to note that solar panels provide peak demand watts, more valuable than middle of the night "baseload" watts. A nuclear plant needs to run night and day, regardless of demand. In other words, building Fermi 3 would increase the need for "dispatchable on-demand" electricity to cover peak periods. Solar panels would reduce this need, and indeed would reduce the need for baseload power.

It's also important to note that installing solar panels adds to the supply of electricity gradually over time, rather than adding nothing for the first ten years followed by the sudden introduction of new generating capacity. If more or less electricity is needed in a particular location than was imagined at the outset, adjustments can be made with solar construction planning. The nuclear reactor can't make that kind of adjustment.

There are also other advantages of numerous solar farms distributed over the service area, compared to a centralized nuclear reactor. The federal Energy Information Administration estimates that national electricity transmission and distribution losses average about 6% of the electricity that is transmitted and distributed in the United States each year. (See http://www.eia.gov/tools/faqs/faq.cfm?id=105&t=3)

Proper planning to site solar farms near demand for peak power would mean less transmission losses. No new high-voltage "transition corridor" leading away from the reactor would be needed. With the sources of power close to peak demand, even lower-voltage distribution line power losses could be cut.

And of course, the fuel cost for solar panels is absolutely zero. We do not have to pay the sun to shine.

The nuclear fuel for Fermi 3 would cost about $100 million per year. $100 million is a bare minimum cost, to which could be added $300 million or more if we attempt to account for the cost of disposal of spent fuel. This added amount is really impossible to quantify, because we still do not know a satisfactory method to dispose of spent nuclear fuel. It's an unsolved problem for future generations. Neither the problem nor the cost exists for solar panels.

There will be some failures in any extensive installation of solar panels. These will be distributed failures; distributed over time and space, fixable on a regular maintenance schedule. A nuclear reactor failure is a centralized failure. At a minimum it is a crisis, even if it only means the electrical output is shut off for a month or two.

And then there is the spectacular failure - the meltdown. A meltdown at Fermi 3 could also cause a meltdown at Fermi 2, and a Fermi 2 meltdown could cascade into the same at Fermi 3. Constructing Fermi 3 next to Fermi 2 makes a meltdown at Fermi 2 more likely, and it makes a meltdown at Fermi 3 more likely as well.

One or both reactors melting down could cause the permanent evacuation of a million or more people while also ruining Lake Erie, Niagra Falls and Lake Ontario with radioactive contamination. Depending on which way the wind is blowing, we may have to write off Toledo or Detroit or both. The events at Fukushima prove conclusively that such a result is possible and that assurances of safety coming from the nuclear industry are worthless.

This is the main (and sane) reason spending our money on solar panels rather than a nuclear reactor makes so much sense. If we need the electricity, we can get it without nuclear fission. The disaster of uncontrolled fission that a nuclear reactor makes possible is not possible with solar power.

If a meltdown in Monroe, Michigan happens, all the fine cost analysis of solar versus nuclear will be meaningless. There will be nobody left in the radioactive contaminated zone to enjoy the benefits of any type of electricity.

That's our real choice. Make the possibility of a Fermi 2 disaster more extensive and more likely, or pursue the alternative. Eliminating the possibility of the Fermi 2 disaster means closing Fermi 2 in addition to never building Fermi 3. Put that way, the right choice is obvious.

Art Myatt