Andasol 1 is Europe’s first parabolic trough solar thermal power station, which went online in Nov 2008. It is located on a high desert site in Granada, Spain, which enjoys a high level of direct insolation – an average of 2,136 kWh / m2 / year. The mirror field — turbine infrastructure can yield a peak electricity generation capacity of 49.9 MWe (20 MWe average, see below). It also has a thermal storage system using molten salt.
The purpose of this post is to consider how one might scale up an Andasol 1 type plant in order to meet a rated power demand for 8,000 hours per year — thereby giving it a capacity factor of ~90%, similar to a baseload coal or nuclear power stations. This is a first attempt to improve the comparisons first given in TCASE 4.
But first, let’s look at the technology and current numbers. Here’s a good summary of its main features:
The Andasol 1 storage system absorbs part of the heat produced in the solar field during the day. A turbine produces electricity using this heat during the night, or when the sky is overcast. This process almost doubles the number of operational hours at the solar thermal power plant per year, the company said.
The heat generated in the solar field will be stored in a molten mixture of 60% sodium nitrate and 40% potassium nitrate. Both substances are used in food production as preservatives and are also used as fertilizer. The storage tank consists of two, 14-meter high tanks with a diameter of 36 meters and a capacity of 28,500 tons of molten salt. During the pumping process from the cold to the hot tank, the molten salt absorbs additional heat at an outlet temperature of approximately 280°C, reaching a temperature of 380°C.
A fully loaded storage system can keep the turbine in operation for 7.5 hours, which means almost 24-hour operation of the power plant in during high sunshine periods.
More technical details, including some useful illustrations of the storage system, can be found here and here. In summary, the solar collectors for the existing plant add up to a total of 510,120 square metres (0.51 km2), consisting of 209,664 mirrors along 312 rows with a total length of 24 km, with 90 kilometres of absorption pipes. The total physical area occupied by the plant (after appropriate collector spacing, and allowing for the storage and turbine housing, etc.) is 1.95 km2. The estimated energy yield is 178 GWh / year (I haven’t seen reports of actual performance data), at a capacity factor of 40.7%, and an average power yield of 10.4 W/m2. It will use 560 million litres/year of fresh water, mostly for cooling the steam circuit, drawn from local ground water (a plant using air cooling would have a lower efficiency and would have to be larger to compensate). The lifespan of the plant is estimated to be 30 — 40 years.
Precise construction costs are hard to come by, but it seems to have been about €300 million ($AUD 500 million). The levelised cost of energy (including the energy storage) is estimated to be 45 c/kWh (in Australian cents) — which is about the size of the Spanish feed-in tariff which is set to run for 25 years. Including its charge for electricity to customers, the maximum cost has been capped at 58 c/kWh.
The crucial data for construction material requirements for Andasol 1 is found in the NEEDS report 2008, “Final report on technical data, costs, and life cycle inventories of solar thermal power plants” – specifically, Table 7.3, page 88. Early in the report (page 28), they calculate costs for a solar thermal power station, located in the Sahara (with better insolation than Spain, but let’s skip this detail for simplicity), generating for 8000 hours per annum — close enough to 90%. They base this on 16 hours storage per day, which they project can be achieved by 2020. The value of 16 seems to be an average number of hours per year, rather than the crucial minimum delivery. Given that the time in winter that is suitable for generating with solar thermal technology is about 5 or 6 hours per day (on clear sunny days), the power station would need to have 18 to 19 hours storage to allow it to have a capacity factor of 90% (excluding bad weather).
The base figures for material requires for the current plant work out to be 1,303 tonnes of concrete and 406 tonnes of steel, and 133 tonnes of glass, per rated MW. To increase the capacity factor from 40% to 90%, one would have to roughly increase the size of the mirror field by a factor of 2.25 (90/40) and the thermal storage facilities by 2.5 (18.5/7.5). The larger mirror field can be rationalised on two fronts: (1) more collecting area is required to recharge the larger volume of storage salts, and (2) the solar multiple for winter will be about twice that of summer.
Let’s use a half-way figure from above — 2.4 — as a scaling constant. This gives 3,127 tonnes of concrete, 974 tonnes of steel, and 300 tonnes of glass per MWe delivered at a 90% capacity factor. Scaled up to the size of an AP-1000 reactor (1,154 MWe at 90 % CF), this is 3.61 million tonnes of concrete, 1.12 million tonnes of steel, and 0.34 million tonnes of glass, with the total plant covering ~101 km2 of desert. By comparison, the reactor would require 0.24 million tonnes of concrete and 0.015 tonnes of steel, and occupy 0.04 km2 of land. So, the comparative solar : nuclear ratios comes out as follows:
Ratio of materials/land requirements, for equivalent solar thermal : nuclear (both calculated at 90 % capacity factor):
Concrete = 15 : 1; Steel = 75 : 1; Land = 2,530 : 1
The conclusion? When energy storage is properly accounted for, the material and land requirements for solar thermal vs nuclear power area appallingly lop-sided. Further, if the solar plant doesn’t end up lasting 40 years, and the AP-1000 lasts 60 years (nearly half of the US reactor fleet is now licensed to run for this long), then the numbers get even more skewed.
Needless to say, for concrete and steel — two of the most carbon-intensive products embedded in any power generation facility — this amounts to a large difference in the embodied energy and associated greenhouse gas emissions of the capital infrastructure. As such, the additional mining required to deliver the limestone and iron ore needed to produce the construction materials for solar thermal versus nuclear, must be set against the mining required for uranium (until Generation IV reactors are standard). Anti-nukes who raise the mining objection against nuclear power can’t have it both ways.
Although I have been careful in my calculations, the above figures are nevertheless a first attempt. As such I’m happy to entertain challenges from commenters, and if these criticisms prove to be right, then I’ll happy adjust my comparative figures accordingly.