THE LIMITS TO SOLAR THERMAL ELECTRICITY.
Conjoint Lecturer, Social Sciences and International Studies, University of New South Wales, Kensington, 2052. F.Trainer@unsw.edu.au Tel. 61 02 93851871.
Accessible evidence on the likely output and costs of trough, central receiver and dish technologies is analysed with a view to assessing the capacity of solar thermal systems to meet electrical demand in mid-winter. In view of the lower performance of troughs in winter most attention is given to Big Dishes and central receivers. General conclusions on output and capital cost are derived. Problems of variability and storage are discussed and are found to set significant difficulties despite the capacity for heat storage. It is concluded that for solar thermal systems to meet a large fraction of anticipated global electricity demand in winter would involve prohibitive capital costs. This study adds to the case that energy and greenhouse problems cannot be solved by renewable energy.
Keywords: Renewable Energy. Solar Thermal Power. Limits to Growth.
The prospects for a high proportion or all of world energy demand to be met by renewable energy sources is likely to depend primarily on the limits to solar thermal supply in winter. Although solar thermal systems are likely to be among the most significant renewable energy contributors they are best suited to the warmest regions and previous studies do not make clear how effective they can be in winter, even in the most favourable locations. The purpose of the following discussion is to consider the capacity of solar thermal systems to meet demand in winter, the variability and storage issues involved, and the probable investment costs.
Unfortunately the data required for confident conclusions on these questions is not readily available. This is partly due to the lack of critical studies in this area, but mainly to the fact that solar thermal developers rarely make publicly available the key data on the performance of commercially operating plants. (Heller, 2010, Blanco, 2010, Mancini, 2010.) However there are theoretical studies and models enabling a useful indicative examination of the issue. At least the following analysis offers an approach which can be reapplied as more satisfactory data becomes available.
The ratio of winter to summer output for troughs is considerable, in the region of 1/4 or 1/5, and much greater than for dishes or central receivers. (Odeh, Behnia and
Morrison, 2003, Fig. 2, Kearney, 1989, Fig. 2, Bockamp et al., 2003, IEEE, 1989, Mills, Morrison and LeLeivre, 2003, Czisch, 2001, NREL, 2010.) This is due to the geometry determining that the angle between sun, reflecting surface and absorber is relatively large much of the time. Primarily for this reason and because of relative capital costs (below) troughs are not likely to be able to make a major contribution to electricity supply in winter.
Lovegrove, Zawedsky and Coventry (2006) claim dishes are in general 50% more efficient than troughs or central receivers. The advantages of dishes are firstly that they can be pointed directly at the sun all through the day and thus avoid the cosine problem which affects trough and central receiver or tower systems and for the former are especially serious in winter. Secondly the high concentration ratios enable higher temperatures than troughs, which make possible more efficient generation of electricity via Stirling engines at the focus of each dish. Thus the heat losses and pumping energy losses involved in transferring heat long distances to a generator are avoided. However as dish-Stirling systems under development do not involve storage they cannot be major contributors in winter.
The main non-Stirling dish initiative is the Australian National University 400 square metre “Big Dish”. (Lovegrove, Zawedsky and Coventy, 2006.) Its annual average solar to electricity efficiency has been estimated at just under 14%, but it is anticipated that this can be raised to 19% in future. The figures given are net of operating energy costs and takes into account transient cloud. However it is not clear what the winter solar to electricity efficiency would be. It is an experimental device and has not been used to provide electricity to the grid over extended periods.
The possibility of using such dishes to collect and store heat is uncertain. Kaneff (1991, 1992) sees this as viable, and says losses can be kept to c. 4 – 5 %. However some European dish developers believe the problems of heat transfer through large numbers of moveable joints and for long distances are daunting. (Heller, 2010.) (Troughs involve long distances but fewer moveable joints and most of the piping length is heated absorber.)
The information given by Lovegrove, and by Kaneff show that the high future solar to electricity efficiency claimed for the Big Dish is due to the assumption of a large scale turbine, and therefore would involve large collection fields and long piping distances and heat losses. Both assumed that the turbine will operate at 35% efficiency, which is 36% greater than the figure for the efficiency of the turbine used in the present Big Dish. If the heat losses, insulation costs, for very large fields are greater than those assumed in the figure for a single Big Dish, as seems likely, then the .19 solar to electricity figure used below will be an over-estimate, or the system capital cost estimate used will be an underestimate.
The ANU solar thermal group is exploring the use of the high temperature achieved by dishes to transform ammonia into nitrogen and hydrogen which can be stored without insulation via processes common in the fertilizer industry, and recombined later to release heat. (Lovegrove, et al., 2004, Kreetz and Lovegrove, 2002, Lovegrove et al., 2012.) They estimate and that the storage in-and-out energy efficiency would be 70%. (This seems to be given as the upper end of a possible range under ideal conditions. Kaneff, 1992, p.143 states the efficiency at 60%.)
The reference situation used in the following discussion is Central Australia, which is likely to be among the best global solar thermal sites. The average mid winter monthly DNI in this location is 5.7 kWh/m2/d.
These figures indicate that in winter with insolation of 5.7 kWh/m2/d (the winter average) and a .19 solar to electricity efficiency, electricity corresponding to a 24 hour flow of 31.6 W/m2 would be generated.
As DNI falls the solar-to-electricity efficiency of dish-Stirling systems falls. (Davenport, undated.) The figures suggest that at the typical maximum winter DNI in Central Australia, c. 730 W/m2, i.e., 27% below peak, output would be 40% below peak, a decrease in solar-electricity efficiency of 13% below the level at peak insolation. This effect will not be taken into account below.
The Big Dish exhibits this effect. Figures 3, 4, 10 and 11 from Siangsukone and Lovegrove (2003) show that on day when DNI is c.1000 W/m2 power output averaged about 38-40 kW, but on a day when DNI was around 800 W/m2 output averaged about 20 kW, (reaching 26 kW late in the reported period.) In other words a 20% fall in DNI from the peak value resulted in a 49% fall in electricity generated. This is somewhat puzzling as it is considerably greater than the fall in the dish-Stirling evidence above. Extrapolation indicates that at 730 W/m2 output would have been 16 kW, 41% lower than at peak insolation.
These will be over estimates of output because there are several factors which reduce gross output, to be discussed below. These factors are likely to be more significant in winter. (On the other hand the lower ambient temperature in winter tends to increase turbine efficiency.)
No evidence has been found on the possible effect of lower DNI on the efficiency of the ammonia dissociation process. In general thermal energy systems are most efficient when temperatures are at their highest. It is likely that an ammonia dissociation system designed to operate best with ideal DNI levels would be significantly less than 70% energy-efficient at 700 W/m2. In Central Australia in winter DNI rises to just above 700 W/m2 for only about 4 hours a day. This will be disregarded in the following discussion.
The 31.6 W/m2 average winter output figure is a theoretically derived estimate and is also likely to be too high in view of observed figures from other systems. It is about 50% higher than the winter output per square metre reported by Davenport for dish-Stirling systems, which are usually regarded as the most efficient solar thermal systems, involve no heat transfer problems and losses, and do not involve the c. 30% loss in the ammonia dissociation process. The figure is about the same as the Van Voorthuysen (2006) general estimate of annual average output for a global solar thermal strategy, which also does not involve the ammonia process and associated 30% reduction in energy available.
However the embodied energy cost of the ammonia storage process appears to be very high. Dunn, Lovegrove and Burgess (2011) report that 28 hour storage of the output from a 10 MW dish would require 162 km of 30 cm diameter steel pipe. (This has been confirmed by personal communications.) The embodied energy cost of this amount of steel would amount to more than 40% of lifetime electricity output. Even if this does not disqualify the process it would make it far less attractive than conventional storage via central receivers.
Unfortunately little or no evidence is publicly available on the actual performance of the few central receiver systems in commercial operation as the operators will not release performance information. However some idea of their probable performance can be based on the figures given in Sargent and Lundy (2003), and the NREL (2010, 2011) Solar Advisor Model. The latter source provides three example cases for a Southern Californian site, where the mid-monthly average DNI is 5.2 kWh/m2day. The average mid winter monthly output corresponds to a continuous flow of 27.7 W/m2, or for a 5.7 kWh/m2 site, 30.4 W/m2.
Factors reducing net solar thermal output.
A full energy accounting would have to include the following thirteen factors which reduce the net energy that could be delivered. Only three will be commented on here.
The available evidence on the life cycle embodied energy cost of solar thermal systems is unsettled and unsatisfactory, especially in view of differing assumptions and the absence of estimates which take into account all “upstream” costs, e.g., the energy needed to produce the steel works that produced the steel used in solar thermal plant construction. These factors can double cost conclusions for steel, and treble those for PV modules. (Lenzen,1999, p. 359, Lenzen, 2009, Lenzen and Dey, 2000, Lenzen and Treloar 2003, Lenzen et al., 2006, Lenzen and Munksgaard, 2002, Crawford, 2011, Crawford, Treloar, and Fuller, 2006.)
The studies reported roughly indicate embodied energy costs that do not take into account upstream factors can be 5 – 11+% of lifetime energy produced. (Dey and Lenzen, 1999, p. 359, Weinriebe, Bonhke and Trieb, 2008, Norton, 1999, and Vant-Hull,1992-3, 2006, Kaneff, 1991, Herendeen, 1988, Lechon, de la Rua and Saezes, 2006, Lenzen, 2009, p. 117.) Dey and Lenzen (1999) state 10.7% for a central receiver. A 10% figure will be assumed here.
Loss in long distance transmission. For transmission via High Voltage DC lines from North Africa or the Middle East to Europe, or from the South West of the US to the North Eastern cities, a considerable loss of energy would occur. Mackay (2008) and Czisch (2004) say this could be 15%. (For similar estimates see also Breyer and Knies, 2009, the NEEDS report, 2009, Trieb, undated, Ummel and Wheeler, 2008, Jacobson and Delucci, 2011, pp.1183-4.) Losses in local substations and distribution, i.e., after the long distance high voltage lines reach urban centres, which might be 7% of energy received, must also be included, suggesting a total loss of 22%. However the total loss figure for combined long distance transmission plus local distribution assumed below will be 15%.
Cooling turbines in the desert regions where solar thermal generation would take place sets the problem of access to sufficient water. Cooling via air avoids this problem but is less efficient, requiring 7% of energy output according to the IEA (2010) and Lovegrove et al. (2012), and imposes a 10% cost increase. These figures will be taken into the accounting below.
The other ten energy loss factors will be listed here without comment. They are, the embodied energy cost of the ammonia heat storage system, plant operating and management energy costs, the embodied energy cost of the long transmission lines and transformers etc., passing clouds (“transients”), down time for repairs, the morning start-up delay, storage loses, reflector washing (including supplying water to desert regions), loss in heat transfer to the power block, and cooling of absorbers by the wind.
Taking into account the losses in transmission and local distribution, embodied energy costs, and dry cooling indicates that the net 24 hour continuous rate of electricity delivered at long distance in winter from Big Dishes with ammonia storage would be around 22.5 W/m2 and 23 W/m2 from central receivers.
Evidence and claims regarding the likely long term future capital costs of solar thermal technologies are scarce and vary considerably, therefore estimates cannot be taken with confidence.
Predictions sometimes assume cost “learning curves” observed in other (selected) engineering fields. However that term might best be confined to improvements in an established technology brought about by increased production scale, plant size, experience and technical advance, whereas dish and central receiver technologies (unlike troughs) are not yet established on a clearly preferable path or scale. For instance central receivers in use are a small fraction of the greatly increased 220 MW scale anticipated for these devices and this will set engineering challenges and choices that have not been encountered to date. According to AETA (2012, p. 92) it is not yet possible to draw conclusions regarding learning rates for solar thermal systems.
Stated cost figures assume construction at relatively convenient locations. However most solar thermal and PV farm-level installations will be in remote desert regions. Lovegrove et al. (2012) say that remote construction will multiply capital costs by 1.3 to 1.4.
A crucial but often overlooked cost factor concerns the relationship between the solar multiple, the collection field and the storage volume. Commonly quoted capital costs per kW of capacity usually assume no storage, or 6 hour storage which is the figure applying to the SAM central receiver models. If solar thermal systems are to be major contributors they will need considerably longer storage capacity, for instance to compensate for the lack of PV input at night. It is not clear what an ideal amount would be to enable an effective combination of renewable input sources to best meet demand at any point in time but it would seem that at least c. 12 hours would be required. The BZE claim (Wright and Hearps, 2010) that 100% of Australian energy could come from renewables assumes 17 hour storage for solar thermal plant.
If a central receiver with a peak rating of 100 MWW is to maintain this output for 6 hours after solar radiation has ceased it will need a solar multiple of around 2, (Hinkley, 2012), i.e., the collection field will need to be twice as big to enable collection and storage of the additional heat energy to run the turbine the extra 6 hours. According to Lovegrove et al. (2012, p. 23) the capital cost of increasing a plant’s capacity to operate from storage from 6 hours to 12 hours adds 30% to the cost of a plant with 6 hour storage. However the figures given in AETA (2012, pp. 87 – 89) state a significantly higher cost of adding storage (to a central receiver) than do those from Lovegrove et al. When their estimates from four to five studies are averaged, adding 6 hour storage to a plant with no storage increases the capital cost 40% to 46%, whereas Lovegrove indicates only a c. 10% rise. If the figures from Lovegrove et al. are taken, extrapolating beyond the 12 hour cost given indicates that 17 hour storage would multiply the cost of plant without storage by 1.6 Thus if AETA had provided figures for 12 and 17 hour storage the multiples would have been markedly higher than those given by Lovegrove et al.
A high cost effect is indicated by the fact that the collection field is the largest cost component of a central receiver, accounting for 39% of plant capital cost according to Hinkley et al., (2012.) This indicates that a plant with 17 hour storage and a solar multiple of c. 4 would add 117% to the cost of one with 6 hour storage.
Proposals giving solar thermal power a significant role rarely if ever take this multiplier into account. For instance the BZE proposal (Wright and Hearps, 2010) involves 17 hour storage but no consideration is given to the possibility that this could more than double the sector’s capital cost.
Easily overlooked is the fact that all future cost estimates assume present materials, construction and energy costs, and in future materials and energy inputs are likely to be considerably more expensive than they are now. Clugston (2010) documents large recent increases.
The following discussion deals only with capital costs and does not include operations and management costs, which are significant additions to total lifetime cost. Lenzen (1999) says for large solar thermal plants these costs could add to almost 20% of plant capital cost. Hearps and McConnell (2011) indicate c. 16%.
The cost of very long distance transmission lines also needs to be taken into account. Solar thermal systems are typically located in deserts a long way from demand and the costs of long distance transmission lines should be added. One line might cater for only three 1000 MW solar thermal power stations, so if solar thermal is to be a major contribution to European electricity supply hundreds would be required. According to Czisch (2004) transmission lines from the Sahara to southern Europe under the Mediterranean Sea would probably add one-third to plant cost. DESERTEC proposals refer only to supply over relatively short distances, such as from Morocco to Spain and from Egypt to Turkey. Supply to Sweden or the UK from the best North African regions, which are towards the East of the Sahara, would involve multiples of those costs.
A confusing factor is that capital costs are stated in terms of dollars to produce 1kW of electricity at peak performance, i.e., in 1000 W/m2 radiation. One NREL Sam example quoted costs $658 million and has a nominal peak capacity of 100 MW, thus is rated at $6,580/kW(e) (peak). However the average annual production rate stated for the site modelled is only 40% of peak capacity. This means the cost of building a plant of this kind would be $16,250 for each 1 kW produced, and more for each kWh delivered at distance after deducting all other energy costs and losses. The capital cost of a plant capable of delivering 1 kW in mid winter would be greater again, and to this should be added the cost of delivery at distance net of embodied energy costs. However the cost of locally sited coal-fired plant capable of delivering 1000 MW in mid-winter, and thus requiring no long distance transmission lines, would be c. $2,000.
According to Sargent and Lundy (2003) the “near term future” cost of solar thermal trough systems will be $(US2003)4,589/kW, or $(A2003)6,556 (using the early 2000s exchange rate.) NREL (2005) states that the 2003 cost for the SEGS systems was $(US2003)7,700/kW which would have corresponded to $(A2003)11,000/kW. Viebahn, Kronshage and Trieb, (2004, p. 20, Table 2 – 3, p. 12) state e5300/kW. The example case given in the NREL, (2010) SAM modelling package states $(US2010)8,243/kW. The average of the six estimates given in AETA (2012, p. 87) is $5,764. According to the figures given by AETA adding 6 hour storage to trough plant increases cost 53% whereas adding it to central receiver plant increases it 42%.
This relatively high cost compared with central receivers below, and the low winter output figures above seem to confirm the above indication that troughs are not likely to be major winter contributors.
Luzzi (2000) estimates that the cost of a Big Dish would be $440,000 but in future could fall to one-third of this figure. Luzzi does not provide derivation or support for the prediction. Lenzen’s references (2009) suggest that there is little evidence available on the issue other than from Luzzi. Fig. 8.3.4, p. 119 on levelised costs, in Lenzen’s review can be read to indicate that the estimated future capital cost would be around 37% of the initial figure, and therefore approximately $(A2000)163,000. Adjusted for inflation this figure corresponds to around $(A2010)240,000.
If it is assumed that Big Dish solar to electricity efficiency is .1.9, DNI in winter is 5.7 kWh/m2/d, peak insolation is 900 W/m2, the ammonia storage process reduces efficiency 30%, and half the electricity delivered is direct and half via ammonia storage, then the foregoing figures yield a future capital cost of $3,764/kWe(p).
However this estimate is based on the 67% capital cost fall Luzzi anticipates which seems to be unduly optimistic in relation to estimates for other solar thermal options. As indicated above, the few recent estimates of US and European construction costs (Hearps and McConnell, 2011) indicate approximately 50% falls and the only recent estimates for Australian conditions, from AEMO (2010) and AETA (2012), anticipate only a 35% fall by 2030 in the capital cost of central receiver systems (not dishes). If Luzzi’s 2000 estimated cost fell by 35% the cost of a Big Dish plus ammonia unit would be approaching twice the estimated future central receiver cost (below.)
However this figure does not include the cost of the ammonia storage system, which it has been explained is likely to be unacceptably high in view of the amount of gas pipe required.
Central receiver costs.
Estimates of probable future central receiver cost vary considerably (also noted by Hinkley, 2012) but support a fairly confident approximate figure. Table 2 sets out estimates mainly from review studies (i.e., not from all individual studies accessed.)
Solar thermal capital cost estimates.
Source. Present cost. Future cost.
Sargent and Lundy (2003) 9,090/kW(p) $3,220/kW(p) by 2020
$(US2003) which =
Hearps and McConnell, (2010) 50% fall by 2050, 35% for
Jacobson and Delucci (2011) $(US)3,082/kW(p) by 2030,
Estimates taken from IEA. ...a 39% fall.
AETA, (2012, Diagrams pp. 68 - 71) Costs fall by about 30%
to 2020, to $4,728, but then do not fall to 2050.
Average adding 3 other
Interest not included.
6 hour storage.
Lovins, (2011) 30% fall by 2025,
no fall to 2050.
Lovegrove, et al.l, (2012) 30% fall by 2025,
no fall to 2050
Hinkley, et al., (2012) $6,494 (3 hour storage,
$8,066 (9 hour storage,
$,7836 (no details.)
Check FUTURE???????????????The last estimate is expected to fall
to $,5675 by 2017.
SAM package (2010, 2011), $(US2010)6,578/kW(p),
Considerably higher current cost figures than average are stated by Viebahn, Kronshage and Trieb, (2004), The NEEDS study (2008), and Nicholson and Lang (2010. Solar Paces (undated) does not expect a fall in capital costs between 2010 and 2030.
The approach taken here uses the NREL figures for (theoretically modelled, not actually observed), present performance and costs, mainly because they provide all the crucial parameters for the one kind of plant regarded as preferable in view of the preceding discussion, i.e., they represent cost, output in winter, and field size for a 100 MW(p), dry cooled central receiver, with 6 hour storage and interest costs included. When the figure for present capital cost, $6580/kW, is reduced by one-third, the most common assumption in Table 1, the cost more or less aligns with those in the future cost column.
The SAM figures are also valuable in enabling estimation to be based on costs per square metre of collection area in relation to annual output, as distinct from being based on nominal peak ratings. The example case given by NREL has just over 1 million square metres of collection field and a capital cost of $658 million, thus the cost per square metre is $658 at present, or $460 for the longer distant future gibven the above reduction assumption.
The winter output per square metre of collection field for a central receiver with 6 hour storage was derived above as 23 W/m2. This means that the capital cost of the plant needed to deliver one Watt net at distance would be $460/23 = $20.
Note that this figure does not include provision for remote area construction or, much more importantly, for more than 6 hour storage. As was discussed above, providing 17 hour storage to enable 24 hour peak output would greatly increase total plant cost, partly because of the additional storage required but more importantly because of the increased solar field size.
Exploring a global renewable energy budget.
The limits to the potential for renewable energy supply depend mainly on variability, discussed below, and the quantities of redundant plant required to cope with this, and the associated investment costs. Trainer (2010a, 2012) explores an approach to the quantitative question in relation to the possibility of a global renewable energy supply. Following is a simplified application of that approach. The assumptions and derivation are transparent enabling reworking with what might be regarded as more appropriate values. It should be noted in advance that this derivation is based on average solar radiation figures and as will be discussed below these result in marked underestimation of the amount of plant and the capital cost of coping with periods of minimal radiation.
Recent estimates (Moriarty and Honnery, 2009, p. 31), indicate that by 2050 world primary energy demand is likely to be in the region of 1000 EJ/y, and final energy demand is likely to be 700 EJ/y although recent price rises and the GFC have reduced growth rates. It will be assumed that electricity remains at the rich world c. 25% of final demand, and that transport remains at c. 33%, but that 60% of transport energy (i.e., 20% of final energy) can be provided via electricity. This would mean 315 EJ/y of electricity would be required. It is likely that in future many more functions will be transferred to electrical drives, but this will be ignored here.
Lenzen’s review of renewable energy technologies (2009, p. 19) reported that due to the integration problems set by intermittency wind is not likely to be able to provide much more than 20% of electricity demand. The figure for PV is probably somewhat higher and 25% will be assumed here. Let us assume for the sake of simplicity that solar thermal systems are to provide the remaining 55% of the 315 EJ/y electricity demand, i.e., 173 EJ/y. Their supply task over a mid-winter month would be 14.4 EJ, corresponding to a continual flow of 5,560 GW.
To meet this demand in winter via central receivers delivering at distance 23 W/m2 would require 242 billion square metres of collection field, and at $460/m2 the cost would be $111 trillion, or $3.7 trillion p.a. assuming 30 year plant lifetime. This is more than 8 times total world annual investment in all forms of energy in the early 2000s. (Birol, 2003, IEA, 2010, Pfuger, 2004.)
This capital cost would be associated with the provision of 55% of the 45% of total energy required in electrical form, i.e., 25% of total energy required.
Also to be included in a budget for a total renewable supply system would be a) the capital costs of the wind and PV components capable of supplying the other 30% of total energy which would be needed in electrical form, i.e., 210 EJ/y, (which is 3.5 times the quantity of electricity presently generated in the world), b) the capital cost of providing the 55% of total energy not in electrical form, c) the cost of long distance transmission lines, d) the increased cost of construction in remote areas, or e) the provision of 12 to 17 hour solar thermal storage.
It seems clear that unless the assumptions and derivations involved in this exercise are grossly mistaken then in terms of quantity and capital cost considerations based on average winter radiation (i.e., not on variability, considered below), and average (not peak) demand, a system meeting world energy demand in winter from renewable sources in which solar thermal played the major role would impose impossibly high capital investment requirements.
The intermittency problem.
The above exercise deals only with the average winter monthly level of solar radiation but the crucial issue for renewable energy technologies is how well they can cope with the intermittency of renewable energy sources and the minimum levels likely. Although solar thermal technologies are in a better position to do this than wind or PV because of their capacity to store heat, an examination of some basic figures indicates that this capacity is not likely to enable solar thermal systems to meet demand through lengthy periods of low radiation, let alone to carry total system loads when wind and PV contributions are low.
The crucial questions concern the variation of DNI around the winter monthly average, the minimum that can occur, and the peak demand that can occur. At Central Australian sites winter DNI averages around 5.7 kWh/m2/d but in particular months it can average 40% below this value, i.e., 3.42 kWh/m2/d. (See ASRDHB 2006, RREDC undated, Hagen and Kaneff 1991, NASA, 2010.) This means that on many days in such a month DNI would be considerably lower than this average.
Peak demand can be 30% above average demand. Therefore to meet peak demand when radiation is 40% below the winter average implies the need for a supply capacity, for instance collection area or storage capacity, some 3.25 times as great as that which would meet average demand at average winter radiation.
An even more difficult problem is yet to be dealt with. This concerns “big gap” events, i.e., protracted periods of calm and cloudy weather, which can cover a continent for a week or more. Some indication of the magnitude of the problem is appropriate here.
Data from the US Dish site at Daggett (Davenport, undated) shows that the 10 lowest days in a December of the reported year averaged only 2.45 kWh/m2/d with a total of only 5 hours over 700 W/m2. For February the average for the 9 lowest days was 2.1 kWh/m2/d with only 4 hours over 700 W/m2 in these 9 days. In December there was one sequence of 5 days of very low DNI and in January there was one of 4 days and another of 5 days, separated by 2 days. In February there was a 4 day run in which there was a total of only 4 hours over 700 W/m2. Note again that at 700 W/m2 dish-Stirling output falls to 50% of peak output.
The DNI data reported by Davenport for a year at Phoenix Arizona are even less favourable than for Daggett. In January the 13 lowest days had an average 2.35 kWh/m2 radiation and these days included only 5 hours over 700 W/m2. At the Mod 2 site over a 19 day period there was dish output on only 2 days, totalling 25 kWh, less than 2% of the level that peak output would have been for that period.
Australian climate data aligns with the above data from the U.S. NASA radiation data shows 21% cloud cover for central Australia in July. At Alice Springs each of the three winter months averages 5 to 8 “cloudy” days and only 17.1 “clear” days. The ASRDHB (2006) provides tables on the probability of sequences of cloudy days at Australian sites. (Table 7, for each location.) At Alice Springs the probability of a 5, 7 or 9 day run in which average daily global radiation in winter is under 4.86 kWh/m2/d is 100% in all cases. (DNI data is not given but other tables show that DNI is around 15% lower than global.) There is a 90% chance of a 4 day run averaging under 3.75 kWh/m2, and in each of the 4 winter months there is a 25% chance of a 4 day run averaging under 3.75 kWh/m2/d. Even on a 4.86 kWh/m2 (global) day DNI would rarely reach 700 W/m2. There is a 50% chance in June that there will be a 5 day run under 4.1 kWh/m2/day of global radiation; i.e., one will occur every two years, and there is a 50% chance that there will be a 7 day run under 4.1 kWh/m2/day. There is a 100% chance that in June there will be a sequence of 14 days in which global radiation is under 5.5 kWh/m2 day. This means DNI would be under 4.8 kWh/m2/d, i.e., under 85% of the 5.7 kWh/m2/d winter average for central Australia. (This has been confirmed with ASRDHB by personal communication.) Elliston (2012) documents “Some very long low irradiance events...” at high quality Australian sites, including four and six day events.
These daily average radiation figures could easily obscure the issue of critical minimum levels. It does not necessarily follow that power generated on a day when DNI totals 2.5 kWh/m2 would be half that generated on a day when it totals 5 kWh/m2. As the above figures show, for little or none of the time during the low day would DNI dries above 700 W/m2, meaning that most of the time there might be no power generated and net plant output would be negative due to “parasitic” consumption of electricity needed to run the plant. This is why Figs. 1 – 3 from Odeh, Behnia and Morrison (2003) shows that below kWh/m trough output plunges.
To summarise, in a typical winter in the best solar thermal regions in Australia it is virtually certain that there will be one 4 day period and several shorter periods in which there is negligible or no generation of electricity, and on almost half of the days DNI would barely reach 700 W/m2.
These observations are not so important for solar thermal researchers and commercial developers because their primary concern is that summer and annual average output reaches economically viable levels. However they are crucial for the discussion of the viability of totally renewable energy supply, which must be maintained in winter and which would probably have to depend greatly on solar thermal sources.
The resulting redundancy issue.
This evidence on protracted periods of cloudy weather shows that the estimation of total system costs should not be based on the cost of the number of solar thermal plants corresponding to average demand, which is a common practice. The typical approach is to estimate a “levelised cost of electricity” (LCOE) produced by a renewable technology taking all life cycle costs and dividing by lifetime output. Near future LCOEs comparable with coal and gas fired generators are often indicated. However these values are highly misleading indicators of the viability of renewable energy systems.
The most important indicators of total energy supply system costs are to do with the amount of redundant plant required within the system to deal with those periods when wind and/or PV and/or solar thermal systems will not be making little or no contribution. If for instance there are long periods of calm and cloudy weather which exceed the capacity of wind, PV and solar thermal systems to deal with then there would have to be sufficient biomass or other supply systems to turn to. In other words total system capital costs would have to include those of the wind, PV and solar thermal etc. components, plus sufficient biomass generating capacity to meet total demand at some points in time.
The magnitude of the problem is illustrated by the proposal by Hart and Jacobson (2011) claiming to show how 2050 Californian electricity supply could come almost entirely from renewable sources. The demand to be met is 66 GW but to ensure that gaps in solar and wind availability are overcome no less than 281 GW of renewable plus gas capacity has to be assumed. The amount of gas fired generating capacity assumed is 75 GW yet it wiould function at a mere 2.6% capacity. Thus an extremely large quantity of redundant plant, four times demand, would have to sit idle much of the time. Lenzen, 2009, notes the need to include these costs of the required ”backup systems” when calculating the real costs of any specific renewable technology, but does not detail the possible magnitude of the problem nor its significant and usually overlooked implications for capital costs.
Note also that apart from time out for breakdowns and maintenance, a coal-fired power station will operate at 100% of peak output whereas a solar thermal power station without storage will only average about 20-25%, because on average the sun shines for only a third of a cloudless 24 hour day. It is therefore more meaningful to attend to capital cost per constant or average delivered kW than to capital cost per kW generated at peak output. Thus a solar thermal power station with a “nominal” peak capital cost that is at present around 3 times that of a coal-fired power station but delivers on average at only .2 of its peak rating has a capital cost per kW delivered that is around 15 times that of the coal fired power station.
The storage issue.
The seriousness of these problems depends on the potential of solar thermal heat energy storage. Some enthusiasts have claimed that this can maintain supply though lengthy calm and cloudy periods, not only keeping output from solar thermal systems up to normal/average levels, but also filling the gaps left by wind and PV systems. It would seem however that large scale long term storage would be highly problematic.
Some solar thermal systems currently operating have the capacity to store for up to 7.5 hours, i.e., to store around 1,500 MWh(th). However if electricity from a 1000 MW(e) solar thermal plant was to be despatched from stored heat for 4 cloudy days, some 280,000 MWh of heat would have to be stored. Heat losses would be considerably greater than the present c. 1%.
According to Hinkley et al. (2012) storage costs $90 per kWh(th), meaning that the cost of four day storage would be some 50% greater than the cost of an entire central receiver with 6 hour storage. (They believe the cost could fall to 25% of this present figure.) According to the NREL Sam cost breakdown (2011), to double storage would be to add some 4 – 8% to plant cost.
Much more important however are the implications for the collection field area. Present cost estimates assume the collection of sufficient energy to deliver through daylight hours plus 6 night time hours, and therefore involve a solar multiple of around 2, i.e., a collection field twice as large as would drive the turbine at peak output through the daylight hours. To enable 24 hour operation via 17 hour storage would require a solar multiple of around 4. According to Hinkley et al. the field cost (presumably for a plant with 6 hour storage) is 39% of total plant cost, so for a plant with a solar multiple of 4 and thus double the amount of field for a plant with 6 hour storage, the additional cost would be 78% of a plant with only 6 hour storage. Thus the cost implications of enabling constant 24 hour supply would seem to be formidable, let alone of enabling big-gap events lasting several days to be overcome.
The solar multiple required to enable four day output from storage might not be markedly greater than that required to enable 24 hour supply, as surpluses might be accumulated gradually. This would depend on how slowly the typical weather regime for the region allowed the storage to be filled, that is, on how often four day or longer operation from storage was called for.
These implications to do with the problem of maintaining output from a single solar thermal plant would seem to clearly rule out the possibility of the solar thermal sector’s combined storage capacity meeting total electricity system demand through protracted periods when there is also little input from wind and PV sectors.
Because of the inaccessibility of basic performance data this analysis has not been able to come to precise or confident conclusions. However the evidence accessed suggests that although solar thermal systems will be valuable contributors they will not be able to make a large contribution to meeting predicted demand levels in winter, let alone to solve the total system problem set by the variability of other renewable sources.
This discussion aligns with previous attempts to assess the potential and the limits of renewable energy sources. (Trainer, 2007, 2010a, 2012.) These conclude that no combination of nuclear, hydro, geo-sequestration and renewable energy sources can sustain energy-intensive societies within safe greenhouse gas emission levels. If this conclusion is sound then the global sustainability problems, centring especially on greenhouse emissions and peak oil, cannot be solved on the supply side.
This has not been an argument against transition to full dependence on renewable energy sources. It has been a contribution to the general case that energy-intensive societies cannot be run solely on these sources. If this is so then in order to solve major global problems it will be necessary to achieve dramatic reductions in rich world per capita resource and energy demand, and a zero-growth economy in which there is a high level of localism and the acceptance of non-affluent lifestyles. Such a transition would be of unprecedented proportions, is quite unlikely to be made, and is not being considered in current policy discussions. That this alternative, “Simpler Way” vision, is workable and attractive is argued in Trainer 2010b, and Trainer 2011. At present the failure to consider such a path derives largely from the common belief that renewable energy sources will eliminate any need to take it seriously.
ABARE, 2010. List of major electricity generation projects. April.
AEMO, (Australian Energy Market Operator),2010. NTNDP Modeling Assumptions: Supply Input Spreadsheets, Available from: http://www.aemo.com.au/planning/ntndp2010consult.html,
AETA, (2012). Australian Energy Technology Assessment. ABARE, Canberra.
Australian Solar Radiation Data Handbook (ASRDHB), 2006. ANZ Solar Energy Society, April, Energy Partners.
Birol, F., 2003. World energy investment outlook to 2030, IEA, Exploration and Production: The Oil & Gas Review, Volume 2.
Blanco, J., 2010. Head of Environmental Applications of Solar Energy, Platforma Sollar de, Almera, Spain. (Personal communication.)
Bockamp, S., Gristop, T., Fruth, M., and Ewert M., 2003. Solar thermal power generation, http://solarec-egypt.com/resources/Solar+Thermal+PowerGen-2003.pdf
Breyer, C and Knies, G., 2009. Global energy supply of concentrating solar power, Proceedings of Solar PACES, Berkeley. Sept, pp. 15 – 18.
Clugston, C., (2010), Increasing Global Nonrenewable Natural Resource Scarcity—An Analysis, The Oil Drum, Apr. 6.
Crawford, R., 2011. Towards a comprehensive approach to zero emissions housing, Architectural Science Review, 54. 4, PP. 277 – 284.
Crawford, R, Treloar, G. J., and Fuller, R., J., 2006. Life cycle energy analysis of building integrated photo voltaic (BIPVs) with heat recovery unit. Renewable and Sustainable Energy Reviews, 10, 559 – 576.
Czisch, G., 2001. Global Renewable energy potential; approaches to its use. http://www.iset.uni-kassel.de/abt/w3-w/folien/magdeb0030901/
Czisch, G., 2004. Least-cost European/Transeuropean electricity supply entirely with renewable energies. www.iset.uni-kassel.de/abt/w3-w/project/Eur-Transeur-El-Sup.pdf
Davenport, R., et al., undated. Operation of second-generation dish/Stirling power systems, Science Applications International Corporation, San Diego.
Dunn, R., K. Lovegrove, and G. Burgess, 2012. A review of Ammonia based thermochemical energy storage for concentrating solar power, Proceedings of the IEEE, 100, 2, Feb ., 391-400
Elliston, B., 2012. Analysing power system impacts using solar radiation data for Australia, http://www.ceem.unsw.edu.au/content/userDocs/presentation.pdf
Hagen, D. L., and S. Kaneff, (1991), Application of Solar Thermal Technology in Reducing Greenhouse Gas Emissions, Dept of arts, Sport, Environment, Journalism, Territories. Canberra.
Hearps, P. and McConnell, D., 2011. Renewable Energy Technology Cost Review. University of Melbourne. http://energy.unimelb.edu.au/index.php?page=technical-publication-series
Heller, P., 2010. Personal communications.
Herendeen R A. Net energy considerations, in Economic Analysis of Solar Energy Systems 1988; MIT Press, Cambridge.
Hinkley, et al., 2012. Concentrating Solar Power Drivers and Opportunities For Cost Competitive Electricity, CSIRO, Canberra
International Energy Agency, 2010. Technology Roadmap: Concentrating Solar Power, International Energy Agency.
IEEE Power Engineering Review, SEGS, 1989. “Solar electric generating stations,” Aug., http://ieeexplore.ieee.org/iel5/39/2848/00087383.pdf?arnumber=87383
Jacobson, M. Z. and Delucci, M., 2011. Providing all global energy with wind, water and solar power. Part II: Reliability, system and transmission costs, and policies. Energy Policy, 39, 1170 – 1190.
Kaneff, S., 1991. Solar Thermal Process Heat and Electricity Generation Performance and Costs for the ANU Big Dish Technology, Energy Research Centre, Research School of Physical and Sciences and Engineering, Australian National University, Canberra.
Kaneff, S, 1992. Mass Utilization of Thermal Energy. Canberra, Energy Research Centre.
Kearney Consulting (2010), Solar Thermal Electricity 2025, http://www.estelasolar.eu/fileadmin/ESTELAdocs/documents/Cost_Roadmap/2010-06%20-%20Solar%20Thermal%20Electricity%202025%20-%20ENG.pdf
Kearney, D., 1989. Solar Electric Generating Stations. JEEC Power Engineering Review, August. 5 – 9.
Kreetz, H and K. Lovegrove, (2002), Exegy analysis of an ammonia synthesis reactor in a solar thermochemical power system”, Solar Energy, 73, 3, Sept, 187 – 194.
Lechon, Y., De la Rua C., and Saez, R. 2006. Life cycle environmental impacts of electricity production by solar thermal technology in Spain. SolarPACES2006, B5-S5.
Lenzen, M., 1999. Greenhouse gas analysis of solar-thermal electricity generation. Solar Energy, 65, 6, pp. 353 – 368.
Lenzen, M., 2009. Current state of development of electricity-generating technologies – a literature review. Integrated Life Cycle Analysis. Dept. of Physics, University of Sydney.
Lenzen, M., and Dey, C., 2000. Truncation error in embodied energy analysis of basic iron and steel products. Energy, 25, pp;. 577 - 585.
Lenzen, M., Dey, C., Hardy, C. and Bilek, M., 2006. Life-Cycle Energy Balance and Greenhouse Gas Emissions of Nuclear Energy in Australia. Report to the Prime Minister's Uranium Mining, Processing and Nuclear Energy Review (UMPNER), http://www.isa.org.usyd.edu.au/publications/documents/ISA_Nuclear_Report.pdf, Sydney, Australia, ISA, University of Sydney.
Lenzen, M. and Munksgaard, J., 2002. “Energy and CO2 life-cycle analysis of wind turbines – review and applications.” Renwewable Energy, 26, 339 -362.
Lenzen, M. and Treloar, G. 2003. Differential convergence of life-cycle inventories toward upstream production layers, implications for life-cycle assessment. Journal of Industrial Ecology, 6, 3-4.
Lovegrove, K., A., Luzzi, I. Solidiani and H. Kreetz, (2004), “Developing ammonia based thermochemical energy storage for dish power plants”, Solar Energy, 76, 1 – 3, Jan-Mar., 331 – 337.
Lovegrove, K., Zawadski, A., and Coventy, J., 2006. Taking the ANU Big Dish to commercialization. Proceedings of ANZSES Annual Conference, Solar 2006.
Lovegrove, K., M. Watt, G. Pollock, J. Wyder and J. Dowse, (2012), Realising the Potential of Concentrating Solar Power in Australia, IT Power.
Lovins, A., 2011. Reinventing Fire.
Luzzi, A. C., 2000. Showcase project; 2 MWe solar thermal demonstration power plant, Proceedings of the 10TH Solar PACES Int. Symposium on Solar Thermal Concentrating Technologies, Sydney.
Mackay, D., 2008. Energy – without the Hot Air. http://www.withouthotair.com/download.html
Mancini, T., 2000. Personal Communications.
Mills, D., G. Morrison and LeLeivre, P., 2003. Multi-tower Line Focus Fresnel Array. Proceedings of ISEG, Mona Kai Resort, Hawaii.
Moriarty, P. and Honery, D., 2009. What energy levels can the earth sustain? Energy Policy, 37, 2469 – 2472.
NASA, 2010. Solar Radiation Data Base. http://eosweb.larc.nasa.gov/cgi-bin/sse/grid.cgi?uid=3030
NEEDS, (New Energy Externalities Developments in Sustainability), 2008. Final Report concentrating Solar Power Plants. http://needs-poroject.org/RSIa/RSIa.pdf
Nicholson, M. and Lang, P., 2010. Zero Carbon Australia – A Strategic Energy Plan: A Critique. http://bravenewclimate.com/2010/08/12/zca2020-critique/
NREL, 2010, 2011. System Advisory Model, (SAM), https://www.nrel.gov/analysis/sam/
Norton, B., 1999. Renewable energy – What is the true cost? Power Engineering Journal, Feb, 6 – 12.
Odeh, S.D., Behnia, M. and Morrison, G.L., 2003. Performance Evaluation of Solar Thermal Electric Generation Systems. Energy Conversion and Management, V44, 2425-2443.
Pfuger, A., 2004. World Energy Investment Outlook. International Energy Authority, Berlin.
RREDC, undated. http://rredc.nrel/gov/solar/old_data/nsrdb/redbook/atlas/serve.cgi
Sargent and Lundy, 2003. Assessment of Parabolic Trough and Power Tower Solar Technology Cost and Performance Forecasts, NREL.
Siangsukone, P., and Lovegrove, K., 2003. Modelling of a 400m2 steam based paraboidal dish concentrator for solar thermal power production. ANZSES 2003 Destination Renewables, Nov.
Solar Paces, undated. Solar Power Tower, http://www.solarpaces.org/CSP_Technology/docs/solar_tower.pdf
Trainer, T., 2007. Renewable Energy Cannot Sustain Consumer Society. Dordrecht, Springer.
Trainer, T., 2010a. Can renewables etc. solve the greenhouse problem? The negative case. Energy Policy, 38, 8, August, 4107 - 4114. http://dx.doi.org/10.1016/j.enpol.2010.03.037
Trainer, T., 2010b. The Transition to a Sustainable and Just World, Sydney, Envirobook.
Trainer, T., 2011. The Simpler Way website, http://ssis.arts.unsw.edu.au/tsw/
Trainer, T., 2012. Can Australia run on renewable energy? The negative case. Energy Policy,
Trieb, F., undated. Trans-Mediterranean Interconnection for Concentrating Solar Power; Final Report, German Aerospace Center (DLR), Institute of Technical Thermodynamics, Section Systems Analysis and Technology Assessment.
Ummel, K and Wheeler, D., 2008. Desert power; The economics of solar thermal electricity for Europe, North Africa and the Middle East. Centre for Global Development, Dec.
Van Voorthuysen, E. H., 2006. Large Scale Concentrating Solar Power (CSP) Technology; Solar Electricity For The Whole World”, in V. Badescu, Cathcart, R. B. and Schuling, R. D. Macro Engineering, Dordrecht, Springer, 2006.
Vant-Hull, L.L., 1992 – 1993. “Solar thermal electricity, an environmentally benign and viable alternative”. Perspectives in Energy, Volume 2, pp. 157 – 166.
Vant-Hull, L. L., 2006. Energy return on investment or solar thermal plants. Solar Today, May/June, pp. 13 – 16.
Viebahn, P., S. Kronshage and Trieb, F., 2004. New Energy Externalities Developments for Sustainability, (NEEDS), Project 502687.
Weinrebe, G., Bonhke, M. and F. Trieb, (2008), Life cycle assessment of an 80 MW SEGS plant and a 30 MW Phoebus power tower, in Proceedings, Solar 98: Renewable Energy For The Americas, ASME International Solar Energy Conference, Alberquerque NM, 13 – 18 June.
Wright, M and P. Hearps, (2010), Australian Sustainable Energy
Zero Carbon Australia Stationary Energy Plan, Energy Research Institute, Melbourne University, Australia.