Wednesday, April 17, 2013

ESOI for solar thermal

Recently, Barnhart & Benson [1] introduced a new metric to evaluate various technologies for energy storage.  They analysed seven storage technologies based on batteries, flow batteries and geologic storage, but did not consider thermal storage. 

The unanswered billion dollar question is how well do solar thermal storage technologies rate on their metric?

The Barnhart-Benson metric, Energy Stored On Invested (ESOI), is the ratio between the energy a device can store in its entire life and the energy required to build the device.

The larger the ESOI, the better is the storage system.  Larger values of ESOI can be obtained by
  • increasing the number of cycles
  • increasing the round-trip efficiency
  • increasing the depth of discharge
  • decreasing the embodied energy
Barnhart & Benson gave the following ESOI values:
compressed air energy storage
pumped hydro storage
Li-ion battery
Sodium-Sulphur battery
Vanadium redox battery
Zinc-Bromine battery
Lead-acid battery

The conclusion by Barnhart & Benson was that

“over their entire life, electrochemical storage technologies only store 2-10 times the amount of energy that was required to build them”.

Clearly that news will not be welcomed by proponents of electrochemical storage. You can bet that feverish work is under way in hundreds of research laboratories around the world to boost the ESOI score.

Published information is available to evaluate the ESOI score for the most common solar thermal storage technology – a molten 60-40 mixture of sodium and potassium nitrates, commonly known as solar salt. 

Burkhardt, Heath and Turchi [2] made a life cycle assessment of a hypothetical 100 MW parabolic trough concentrating solar plant at Daggett, California.  The storage envisaged is 62,000 t of solar salt, capable of storing 1,988 MWh of thermal energy, which can be converted into an electrical equivalent by multiplying by the thermal-electric efficiency of the plant.

Many individual items were taken into account by Burkhardt et al. to calculate the embodied energy of the storage component of the plant; these included obvious items like steel, concrete, pumps, heat exchangers, insulation and solar salt.  However the biggest single item is the energy required to keep the salt molten and stirred for daily operations.

It’s noteworthy that the embodied energy of solar salt is low if it mined (as assumed to be the case in [2]), but high if it produced synthetically.  In the latter case, which Burkhardt et al. say applies to slightly more than half of all installations, the manufacturing process involves pre-production of ammonia, for which there is a natural gas requirement.

I have also made an as-yet unpublished estimate for the ESOI score for thermal storage in air-blown pebble beds.  This estimate is in the context of a new concept for solar thermal power generation entitled BRRIMS, denoting Brayton-cycle, Re-heated, Recuperated, Integrated, Modular and Storage-equipped.  Here what needs to be considered is the embodied energy in hardware such as steel tanks, ducts, concrete footings, insulation and pebbles.  Heat exchangers, pumps and fans are not required.

Results of Barnhart & Benson can now be extended as follows, with the new data highlighted.  This is a fair comparison (“apples with apples”) between storage technologies since the new figures represent electrical energy that would be produced from the underlying thermal storage. 

compressed air energy storage
pumped hydro storage
pebble bed thermal, BRRIMS
solar salt, parabolic trough [2]
Li-ion battery
Sodium-Sulphur battery
Vanadium redox battery
Zinc-Bromine battery
Lead-acid battery

The simple conclusion from the ESOI metric is that geologic storage is excellent, thermal storage is good, whilst electrochemical storage is poor.

That is not the whole story however.  Geological storage is not particularly cheap, and its applicability is limited by the availability of suitable sites.  My estimates show that thermal storage is the cheapest option, and I propose to present details of this work at the World Renewable Energy Congress in July.


[1] C J Barnhart and S M Benson, “On the importance of reducing the energetic and material demands of electrical energy storage”, Energy Environ. Sci., 6 (2013), 1083.

[2] J J Burkhardt III, G A Heath and C S. Turchi, “Life cycle assessment of a parabolic trough concentrating solar power plant and the impacts of key design alternatives”, Environ. Sci. Technol. 45 (2011), 24572464.


Friday, April 5, 2013

Cost of solar power (34)

Usually, a student of solar power does not have access to high-quality information about projects.  Financial information is almost always lacking, reasonably enough since it’s commercially sensitive.  In other cases, project specifications are given in headline form only, direct from the Media Release, without concern for engineering accuracy.

But sometimes – rarely, once in a blue moon – all necessary information is readily available, as discussed in this post.

Three years ago, the National Renewable Energy Laboratory (NREL) published a report [1] on the design of a 100 MW parabolic trough solar thermal power station in Daggett, California.  The design includes 6 hours molten salt thermal storage.  The power block is conventional steam Rankine-cycle with wet cooling, although analysis of dry cooling was included as an option.

I should mention that this is a design exercise, not a plant that has been constructed.

NREL appointed consulting engineers, Worley Parsons, to design the plant.  And NREL is to be congratulated because the 112 page report includes extremely comprehensive details, both financial and engineering.  I recommend this report strongly if you wish to understand the ins-and-outs of solar thermal power.

In brief, the Daggett specifications envisage 100 MW_e power output.  There are 405,888 parabolic mirrors with area 987,540 m^2, 43,488 solar collector receiver tubes and 1,208 solar collector assemblies/drives.  The plant footprint is about 4 km^2, and the thermal storage of 1,988 MWh_th requires 62,000 tonnes of solar salt (a 60-40 mix of sodium and potassium nitrates).  The annual output is 426,717 MWh_e and the cost of the plant in 2010 USD is $1.016 billion.

To give an indication of the quality of the information in the report, the parasitic power losses with the wet-cooled plant are given explicitly.  The gross output of the steam turbine is 118 MW_e, the pumps to circulate the Heat Transfer Fluid (HTF, a synthetic oil) consume 7.9 MW_e, the condensers consume 2.0 MW_e and other parasitic losses are about 5 MW_e. 

As a further indication of the detailed nature of the report, here are the listed components of the thermal storage system:
  • 2 cold salt tanks (including foundations)
  • 4 cold tank immersion heaters
  • 2 hot salt tanks (including foundations)
  • 4 hot tank immersion heaters
  • 3 hot salt pumps
  • 3 cold salt pumps
  • 6 salt to HTF heat exchangers
  • 62,000 t bulk salt storage
  • and a nitrogen storage and vaporisation system, especially to minimise fire risk should the HTF vaporise
Well, I could go on at length, but hopefully the point is clear.  This report is a goldmine for anyone who wants to know the constituent parts of a parabolic trough solar thermal power station.

Let me now turn to the main point of this series of blog posts – the cost of the solar power that would be produced by the Daggett plant were it to be built.  The report gives a figure for the Levelised Cost of Electricity (LCOE), namely USD 184/MWh_e “nominal”, which I take to mean before inflation.  However this figure is based on an Income Tax Credit, which is not available outside the USA, so I’m going to calculate the LCOE using my standard methodology, for which the assumptions are:
  • there is no inflation,
  • taxation implications are neglected,
  • projects are funded entirely by debt,
  • all projects have the same interest rate (8%) and payback period (25 years), which means that the required rate of capital return is 9.4%,
  • all projects have the same annual maintenance and operating costs (2% of the total project cost), and
  • government subsidies are neglected.

For further commentary on my LCOE methodology, see posts on Real cost of coal-fired power, LEC – the accountant’s view, Cost of solar power (10) and (especially) Yet more on LEC.  Note that I am now using annual maintenance costs of 2% rather than 3% as in posts during 2011.

The results for the Daggett design are as follows:

Cost per peak Watt              USD 8.95/Wp
LCOE                                     USD 272/MWhr

The components of the LCOE are:
Capital           {0.094 × USD 1,016×10^6}/{426,717 MWhr} = USD 224/MWhr
O&M              {0.020 × USD 1,016×10^6}/{426,717 MWhr} = USD 48/MWhr

By way of comparison, LCOE figures (in appropriate currency per MWhr) for all projects I’ve investigated are given below.  The number in brackets is the reference to the blog post, all of which appear in my index of posts with the title “Cost of solar power ([number])”:

(2)        AUD 183 (Nyngan, Australia, PV)
(3)        EUR 503 (Olmedilla, Spain, PV, 2008)
(3)        EUR 188 (Andasol I, Spain, trough, 2009)
(4)        AUD 236 (Greenough, Australia, PV)
(5)        AUD 397 (Solar Oasis, Australia, dish, 2014?)
(6)        USD 163 (Lazio, Italy, PV)
(7)        AUD 271 (Kogan Creek, Australia, CLFR pre-heat, 2012?)
(8)        USD 228 (New Mexico, CdTe thin film PV, 2011)
(9)        EUR 200 (Ibersol, Spain, trough, 2011)
(10)      USD 231 (Ivanpah, California, tower, 2013?)
(11)      CAD 409 (Stardale, Canada, PV, 2012)
(12)      USD 290 (Blythe, California, trough, 2012?)
(13)      AUD 285 (Solar Dawn, Australia, CLFR, 2013?)
(14)      AUD 263 (Moree Solar Farm, Australia, single-axis PV, 2013?)
(15)      EUR 350 (Lieberose, Germany, thin-film PV, 2009)
(16)      EUR 300 (Gemasolar, Spain, tower, 2011)
(17)      EUR 228 (Meuro, Germany, crystalline PV, 2012)
(18)      USD 204 (Crescent Dunes, USA, tower, 2013)
(19)      AUD 316 (University of Queensland, fixed PV, 2011)
(20)      EUR 241 (Ait Baha, Morocco, 1-axis solar thermal, 2012)
(21)      EUR 227 (Shivajinagar Sakri, India, PV, 2012)
(22)      JPY 36,076 (Kagoshima, Kyushu, Japan, PV, start July 2012)
(23)      AUD 249 (NEXTDC, Port Melbourne, PV, Q2 2012)
(24)      USD 319 (Maryland Solar Farm, thin-film PV, Q4 2012)
(25)      EUR 207 (GERO Solarpark, Germany, PV, May 2012)
(26)      AUD 259 (Kamberra Winery, Australia, PV, June 2012)
(27)      EUR 105 (Calera y Chozas, PV, Q4 2012)
(28)      AUD 245 (Nyngan and Broken Hill, thin film PV, end 2014?)
(29)      AUD 342 (City of Sydney, multiple sites, PV, 2012)
(30)      AUD 281 (Uterne, PV, single-axis tracking, 2011)
(31)      JPY 31,448 (Oita, PV?, Japan, to open March 2014)
(32)      USD 342 (Shams, Abu Dhabi, trough, to open early 2013)
(34)      USD 272 (Daggett, California, trough with storage, designed 2010)


The LCOE for the Daggett design is comparable to the figure for Blythe, see Cost of Solar Power (12).  (The Blythe plant has an interesting history; the figures used in my analysis were for a parabolic trough plant to be built by Solar Millenium.  However at a very late stage, as discussed here, the proponents switched the design from solar thermal to PV.) 

The Daggett LCOE is about 30% more expensive than the Crescent Dunes project, see Cost of Solar Power (18), which is now nearing completion.  I think this is mainly due to a general decrease in costs as we proceed along the “experience curve”, perhaps also to an advantage that heliostat/tower plants have over parabolic trough plants, notably in that parasitic losses to pump the HTF around the plant are avoided.

The other important point about the Daggett study is just how comprehensive it is.  I regard it as valuable reading for students of solar power.


 [1]  C. Turchi, “Parabolic trough reference plant for cost modeling with the solar advisor model (SAM)”, Technical Report NREL/TP-550-47605 (July 2010).