A review on life cycle assessment of concentrating solar energy technologies

In the reviewed case studies, different configurations have been assessed for their life cycle impacts (as specified in Table SI1). These configurations are both commercial and non-commercial systems, and the commercialization of the systems does not affect the analysis of their life cycle environmental impacts and costs. Considering the types of receivers, parabolic trough and solar tower are the most common receivers, with 13 papers studying parabolic trough, 12 papers studying solar tower, and five papers studying both types of receivers. This is reasonable as most of global CSP plants are based on either parabolic trough and solar tower, in which 70 % of currently operating CSPs are based on parabolic trough and 20 % are based on solar tower, with installed capacities of 4.46 GW and 1.31 GW, respectively [30,31].

In term of HTF or working fluid, water is normally used in conventional configurations for heating purpose, or to supply hot water. This conventional design may not require a lot of materials such as the concentrating solar system in Ref. [32] which is combined with solar PV panels, uses water as the working fluid, and there is no need for chemicals for HTF or TES. In this system, solar energy is collected by parabolic trough mirrors and reflective panels. The solar heat is transferred by a system of water tubes. This plant includes a support structure and a solar tracking system, but does not include TES [32]. Some other CPV systems which utilized water as the working fluid [33,34], was integrated with a tank to store hot water and PV modules to generate electricity.

At the same time, a concentrating solar system may utilize synthetic oil/fluid/salts or molten salts as HTFs and TES materials to increase the energy efficiency of the system. More than half of the reviewed case studies were conducted on CSP (or CST) plants with molten salts as either HTFs or TES materials. An example of this system is the solar tower plant, being cooled by air and integrated with a thermal energy storage which is assessed in Ref. [35]. Two types of HTFs have been studied, including mined nitrate salts and synthetic salts [35].

The most common TES is two-tank system including a hot and a cold tank, at different storage capacity. The thermocline tank is less common. For power engine, the most common one is the steam engine, Rankine cycle. Recent studies were conducted on CSP plants based on supercritical CO₂ or CO₂-SO₂ blended, with recompression cycle and Brayton cycle [36,37].

Sometimes, a concentrating solar system is combined with other technologies to maximize the solar energy potential and increase the energy efficiency. One example is the CPV technology which utilizes CST or CSP systems to concentrate the solar radiation and the PV cells to capture the solar irradiation. Common concentrating solar receivers are parabolic troughs and lens, while the most popular PV cells are crystalline silicon or thin films. The technical performance of hybrid CSP and PV plants were studied in the literature, indicating some evidence on the improvement of their performance compared to CSP alone. Vossier et al. studied the use of photovoltaic mirror acting as solar cell and reflector, and found that this system generated 50 %–130 % more electricity output compared to CSP [38]. Another study on THEMIS CSP plant in France indicated that both the daily or annual performance of hybrid system are higher than those of conventional plant, with energy gain from 20 % to 55 % higher, depending on pessimistic or optimistic scenario [39]. However, the choice of materials for balancing the electricity conversion of PV and CSP in different temperature conditions should be further studied [38]. In practice, the combination of CSP and PV is the technology development trend at commercial scale, as 7 out of 10 CSPs being under construction globally, are co-located with PV plants, with the total capacities of 1.16 GW of CSP and 2.09 GW of PV [30,31].

In the reviewed case studies, the hybrid configuration is available for both low and high concentration ratios. At a low concentrating ratio, CPV technologies are applied or integrated into a building. These CPV technologies increase conversion efficiency [40,41], or aims at generating electricity to provide the illumination function [42], or providing both electricity and hot water [34].

Another option is to integrate an auxiliary boiler into the concentrating solar system [47]. The purposes of this boiler are to provide heat for starting up the system, to avoid freezing the HTFs, and to generate additional steam and heat. Both fossil fuels (natural gas (NG)) and biomass-based fuels such as biomethane, biogas and biowaste can be used to operate the boiler [[47], [48], [49], [50]]. An example of NG-fueled boiler can be found in Ref. [48], in which a NG boiler, with different percentages of NG ranging from 0 %, 5 %, 10 %, 15 %, 20 %, 25 %, 30 %, and 35 %, is applied in a parabolic trough CSP plant. Other studies compared the life cycle environmental impacts of CSP plants with NG, biogas provided by a nearby plant and biomethane produced from mixed sources of biowaste, sewage sludge, grass, and animal manure [47], and solar only CSP and hybridized CSP with 12 % of electricity generated by NG, coal, fuel oil, wheat straw, wood pellets and biogas [50].

In more advanced systems, heat recovery system (HRS) is included in CSP plants to further increase efficiency [49,51,52]. In these plants, being called hysol CSP plants, solar radiation reflected by the 9151 heliostats is directed to an external central receiver. The solar thermal energy is then transferred by the nitrate molten salt and stored in the hot tanks of the thermal energy storage. The heat is used to drive the Rankine cycle engine to generate electricity. The remaining heat from the steam engine is stored in the cold tank of the thermal energy storage and transferred to the HRS. The purpose of the HRS is to reheat the HTF which will be transferred back to the hot tank of the thermal energy storage, starting a new cycle [49,51,52]. Thanks to the reusing and recycling of thermal energy, the inclusion of HRS helps to improve the efficiency of the system. Besides, the dispatchability of concentrating solar technologies is enhanced by increasing their storage capacity.

The CSP is integrated with the steam-forming reactor to generate electricity and produce enriched methane for fuel [53]. This hybrid plant consists of a parabolic trough as a receiver, molten salts as HTF and a TES. The heat from hot molten salts is used to run the steam reforming system, while the remaining heat is used to generate electricity. At the end of the process, both enriched methane and electricity can be obtained [53].

The life cycle energy consumptions of concentrating solar technologies in the reviewed papers are assessed in terms of Cumulative Energy Demand (CED), Global Energy Requirement (GER), Primary Energy Demand (PED), Fossil Fuel Depletion Potential (FFDP), Energy Payback Time (EPBT), and Energy Return on Investment (EROI). Among these indicators, EPBT and CED are the most common ones, being quantified in 15 and 14 papers, respectively. These two energy indicators are recommended by IEA in conducting LCAs of solar technologies [76,77]. Other indicators such as EROI, FFDP and GER are assessed in one or two papers.

CED is measured in MJ of oil equivalent of primary energy used by the technology during its life cycle [78]. CED can be quantified according to renewable (including hydro, solar wind, geothermal and biomass) or non-renewable (including fossil and nuclear) energy, which are respectively called CED, renewable and CED, non-renewable. GER or PED denotes the same meaning as CED, indicating the primary energy consumption per functional unit of the technology [33]. These indicators were used in the early version of the EPD LCA method and CML method, respectively. The equation for calculating CED (GER or PED) can be found in Ref. [78].

It should be noted that there are several other parameters that can significantly the CED (as well as other environmental indicators) per kWh, such as the location, DNI, technology lifetime, temporal impact. In fact, the correlation between these parameters and CED in the reviewed case studies is not strong. For the location and DNI which are closely related, the correlation is not strong, as most of the reviewed case studies are conducted on CSP plants in high DNI locations (Refer to the life cycle assessment methodological aspects for the location of the reviewed case studies). The technology lifetime is from 20 to 30 years for all case studies, therefore causes no explicit impacts on the CED. The temporal impact may have effect on the LCA database as well as the development of technologies, with the assumptions that the more recent databases are more comprehensive, and the more recent technologies are more efficient. However, the relation between temporal parameter and the CED is not obvious (as can be seen in Figure SI1). This weak correlation can be explained by the small number of datasets, e.g. number of LCA case studies concerning CED per kWh of electricity.

There are studies reporting the CED of concentrating solar technology per less common FUs. For example [53], quantified the CSP combined with steam reforming and reported the CED at 5.97E+01 MJ per one Nm³ of enriched methane and 0.58 kW h of electricity. In Ref. [33], the CED of a CPV system for both electricity and thermal purposes is 2,56E+04 MJ per system, and that of a dish Stirling CST plant in Ref. [60] ranges from 7,21E+03 to 2,04E+04 kW h per system, depending on reuse/recycling or landfill waste management scenarios. The absolute CED values of these systems are higher compared to other case studies because they are calculated per system (1 kW electricity and heating water in Ref. [33] and 1 kW electricity and 2 kW thermal in Ref. [60]), while other case studies reported CED per kWh. Besides, both these studies reported global energy requirement (GER), which are very similar to CED. The only difference between GER and CED is that the first is based on higher heating value, while the latter may be based on either lower or higher heating value.

EPBT is the time period needed for a renewable energy system, in this case, a concentrating solar system to generate the same amount of primary energy equivalent which was used to produce the system itself. EPBT is calculated by the following equation [76]:(1)In which:

The EPBT of concentrating solar technologies ranges from 0.33 to 10.7 years, depending on the specific systems (as presented in Fig. 6b). The lowest EPBT is at 0.33 years for a solar tower CSP plant using thermochemical storage [70]. Some concentrating solar systems have low EPBT (less than one year), such as hysol CSP biomethane thanks to the inclusion of the heat recovery system and use of renewable energy, and CPV systems thanks to the maximum exploitation of solar radiation with both concentrating solar and solar PV technologies. The EPBTs of hysol biomethane are around 0.5 years [49]. Those of CPV systems are from 0.7 years to nearly one year, with the lower end from CPV systems using parabolic trough as receiver [33,44], and the higher end from CPV systems with Fresnel lens or achromatic lens [44], and flat coated reflectors for concentrating system combined with PV and copper cooling pipe [79]. The EPBT of an asymmetric compound parabolic concentrator combined with PV cells ranges from 2.82 to 4.74 years depending on installation locations in China [75]. There were two exceptional CPV with EPBT longer than six years, which were installed in the UK and Israel [41,63].

Most of CSP and CST technologies have the EPBT between 1 and 6.6 years [32,35,42,[48], [49], [50],54,[56], [57], [58],60,69]. There is only one system with EBPT longer than 10 years, which is a dish Stirling CST with landfilling as the end of life treatment [60]. By types of receivers, the EPBT of parabolic trough CSP range from 1 year to 3.43 years per kWh, which are quite similar to that of solar tower CSP, between 0.3 and 3.58 years. The EPBT of solar tower CSP is lower than 1 years for CSP plant with thermochemical energy storage system or with heat recovery system and biomethane backup. Fresnel lens have the EPBT of less than 1 year, for low concentrating PV systems.

EROI considers the relation between the amount of energy produced and the amount of energy consumed during the life cycle of the technology [78]. This indicator was considered in three paper, which range from 3.77 to 6 for the CPV integrated in a building in the UK and EU [[40], [41], [42]].

Numeric results related to energy indicators of different concentrating solar technologies are provided in Table SI2.

The GHG emissions of concentrating solar technologies range from 2.09e+03 to 2.28e+03 kgCO₂eq per kW of installed capacity of the system from cradle to grave. The lower end is high CPV which combines the Fresnel lens with GaInP/GaInAs/Ge multijunctional PV cell [43], while the higher end is CPV combining parabolic trough and crystalline silicon PV cell in Ref. [33]. At the same time, the scales of the two systems are different, in which the system in Ref. [33] is small scale, at 1 kW, and that of [43] is 53 kWp. It should be noted that the original functional unit of both studies is “the system” with different installed capacities, therefore the function unit is converted into kW for harmonization, and convenient for comparison.

Cucumo et al. reported the GWP of a dish Stirling system at 3.55e+03 kgCO₂eq per system, which has installed capacity of 1 kW electricity and 2 kW thermal energy [60]. Besides, Piemonte et al. reported the GWP of a CSP plant at 4.98e-01 kgCO₂eq per functional unit of 1 Nm³ of enriched methane with 17 % hydrogen and 0.58 kW h of electricity [53].

Disregarding CPV or CSP technologies, the lower end of this range is the GWP of CSP considering the manufacturing and operation stage only, 9.8e-03 kgCO₂eq [55]. From cradle to gate, the life cycle emissions are 3.02e-02, 3.70e-02, and 1.90e-01 kgCO₂eq for different systems [35,53,54]. From cradle to grave, the range is from 1.64e-02 to 3.92e-01 kgCO₂eq per kWh. The lowest GHG emissions are from high concentrating PV technologies which combine multijunction PV cells (GaInP GaAs GaInAs Ge) with different concentrating mirrors or lens. This ranges from 1.64e-02 for parabolic mirrors, 1.69e-02 for Fresnel lens and 1.84e-02 for Achromatic lens kgCO₂eq per kWh [44]. The highest GHG emissions are hybrid CSP with fossil fuels. This ranges from 3.11e-01 to 3.92e-01 kgCO₂eq per kWh of electricity generated by solar tower CSP hybrid with natural gas (NG) (35 %) using synthetic oil as HTF [48] and by parabolic trough hybrid with NG (54 %) turbine using molten nitrate salts as HTF [51], respectively. The mean value is 9.27e-02 kgCO₂eq per kWh and the median value is 3.9e-02 kgCO₂eq per kWh.

Similar to EPBT, GPBT denotes the time period a renewable energy system takes to compensate for the GHG emitted during the life cycle of the system. In the reviewed case studies, only CPV technologies were assessed with GPBT, and there was no information on GPBT of CSP and CST. The GPBT of CPV ranges from 0.98 to 2.38 years. The shortest GPBT is 0.98–1.1 years for the high CPV [44]. The report GPBT for the building applied CPV in Italy is 1 year [33]. Other research reported higher GPBT, for example, 1.89 years in Spain and 2.38 years in France for the building integrated CPV [42].

Land use of concentrating solar technologies in the reviewed case studies ranges from 4.00E-06 to 1.30E-03 m² per kWh. The lower end comes from parabolic trough CSP solar only in Spain, at around 4.00E-06 to 5.00E-06 m² per kWh, followed by hybrid CSP with different fuels such as NG and biomass, at around 8.00E-06 to 5.01E-05 m² per kWh [47,48,50]. The higher end comes from CSP in the USA, ranging from 2.30E-04 to 1.30E-03 m² per kWh [62,73]. The average land use of these technologies is 1.78E-04 m² per kWh and the median value is 1.00E-05 m² per kWh.

Considering that the technical characteristics of CSP plants in Corona et al.’s studies those in the USA of Klein and Rubin [62] and Rangarajan et al. [73] are very similar in term of electricity generation efficiency, at around 80 GW h per MW of installed capacity over their life cycle, the large difference among the land use of these CSP plants is likely due to the assumptions of LCA calculations. While Corona et al.’ studies applied the ReCiPe for LCIA method and Ecoinvent data for the background processes, Klein and Rubin [62] used hybrid LCA combining process based and economic input output methods. For land use, both studies on CSP plants in the USA quantified the direct land use of the plants over their 30-year lifetime.

Water requirements for wet cooled CSP plants are from 3.69E+00 to 7.10E+00 m³ per MWh, and those of CSP plants using synthetic HTFs are from 6.24E+00 to 6.41E+00 m³ per MWh. The highest water requirements are for a wet-cooled parabolic trough CSP in the USA (7.10E+00 m³ per MWh) [62] and for a hybrid CSP and biomass using synthetic oil as HTF (6.41 E+00 m³ per MWh) [47].

The water consumption in wet cooling CSP plants is five times higher than that of dry cooling plants, at average. Considering that most of the water is consumed during the operation stage, for cooling system in the power generation unit and cleaning the mirrors, it is reasonable that the types of cooling system have a considerable influence on the water consumption. Meanwhile, the type of HTF has less impact on water consumption. The average water consumption in CSP plants with synthetic salts as HTFs is from 1.4 to 1.8 times higher than those using mined salts as HTFs, for dry cooling and wet cooling system, respectively. In the same study, the difference is even smaller, at around 4 %, e.g. 4.9 and 4.7 m³ of water per MWh [58]. This difference must originate from the water consumption for background processes of HTF production.

It is expected that the cofiring of biomass in CSP plant will increase the water requirement over the life cycle, due to the water consumption for both biomass production and water for cooling. In fact, the use of biomass as the co-firing fuels does not always impose the large amount of consumed water. This can be seen from Fig. 8 that the results are contradictory among water requirement of CSPs used biomass for co-firing or auxiliary purposes. Several biomass related CSP plants (with ‘cofiring BM‘ in the technologies’ names, in the ‘y’ coordinate of Fig. 8) have either considerable or insignificant water requirements.

The material requirement of concentrating solar technologies is presented in abiotic depletion potential, which is from 1.93E-05 to 1.60E+00 g Sb eq per kWh. The abiotic depletion is extremely low for hybrid CSP and biomass in Tusinia, at 1.93E-05 g Sb eq per kWh [71], followed by CSP with micro gas turbine in Italy, from 2.85E-04 to 1.28E-03 g Sb eq per kWh [72]. Other technologies such as solar tower cofiring with fossil fuels, solar tower CSP, parabolic CSP and dish Stirling CSP impose a higher material requirement, from 2.02E-02 to 1.60E+00 g Sb eq per kWh [56,57,67,74]. The average material requirement of CSP is 3.56E-01 g Sb eq per kWh and the median value is 2.20E-02 g Sb eq per kWh.

The acidification potential of concentrating solar technologies ranges from 1.40E-01 to 9.80E+00 g SO₂ eq per kWh, disregarding the system boundaries, with the median value at 3.63E-01 g SO₂ eq per kWh. The acidification potential largely depends on the types of fuels used for auxiliary purposes or co-generation. The larger amount of SO₂-intensive fuels such as NG and biomass will increase the acidification potential. CSP with biomass backup burner cause the largest acidification potential impacts, at 9.80E+00 [59] g SO₂ eq per kWh. Other technologies such as solar tower CSP cofiring with 12 % of diesel (CSP47a in Fig. 9a), hybrid CSP with NG (CSP21b), hysol CSP with NG (CSP21c), cause relatively large acidification potential, at 1.34E+00 g SO₂ eq per kWh [67], 6.94E-01, and 8.91E-01 g SO₂ eq per kWh [51], respectively. The lower range of acidification potential comes from solar tower CSP without cofiring (CSP47c in Fig. 9a), at 1.40E-01 g SO₂ eq per kWh [67] and parabolic trough CSP solar only, using NG as auxiliary fuel (CSP46a, CSP19a, CSP20, CSP19c, CSP19b), at around 1.66E-01 and 1.68E-01 g SO₂ eq per kWh [47,48,50]. Considering the cradle to grave system boundary, the acidification potential of concentrating solar technologies narrows down to 1.66E-01 to 8.91E-01 g SO₂ eq per kWh, with the mean value of 4.01E-01 g SO₂ eq and the median value of 3.62E-01 g SO₂ eq per kWh.

The eutrophication potential of concentrating solar technologies is quantified in either g P eq or g PO₄ eq. For the harmonization purpose, the unit of gPO₄ eq will be converted into g P eq based on their atomic weight. The eutrophication potential considerably ranges from 4.33E-09 to 1.79E+01 g P eq per kWh. For all the cases of CSPs in Spain and other countries conducted by Corona et al. [48,[50], [51], [52],80], the eutrophication potentials are in the high range from 8.09E+00 to 1.79E+01 g P eq per kWh, with the mean value of 1.27E+01 and the median value of 1.18E+01 g P eq per kWh. Four case studies have the lower eutrophication potential from 3.26E-03 to 3.26E-01 g P eq per kWh [56,57,59,67], with the average eutrophication potential at 6.84E-02 g P eq per kWh. The lowest eutrophication potential comes from small scaled CSPs, from 4.33E-09 to 3.62E-05 g P eq per kWh [71,72], with the average eutrophication potential at 6.03E-06 g P eq per kWh. Due to the discrepancy in the eutrophication potential in the existing case studies, no clear conclusion can be drawn, except for that the scales of CSP, may have impacts on their eutrophication potential.

Ecotoxicity includes freshwater ecotoxicity, marine ecotoxicity and terrestrial ecotoxicity, which are quantified in grams of 1,4-Dichlorobenzene equivalent (g1,4-DBeq or g1,4-DCBeq). There is a large difference in the quantified ecotoxicity. The freshwater ecotoxicity ranges from 3.29E-01 to 1.36E+01 g1,4-DCBeq per kWh, with the lower range from parabolic trough CSP solar only [50] and hysol CSP [52]. The terrestrial ecotoxicity ranges from 2.66E-01 to 7.90E-01 g1,4-DCBeq per kWh, being calculated for the solar tower, parabolic trough CSP, dish Stirling CST and the solar tower with various percent of cofiring [56,57,67]. The marine ecotoxicity significantly ranges from 2.08E-01 to 1.15E+05 g1,4-DCBeq per kWh. The higher end of marine ecotoxicity compared to freshwater and terrestrial ecotoxicity suggests that most of the ecotoxicity impact of concentrating solar technology is due to the impacts on the marine ecosystem.

Similar to ecotoxicity, human toxicity is quantified in g1,4-DCBeq. The human toxicity of concentrating solar technologies ranges from 1.07E+01 to 1.26E+02 g1,4-DCBeq per kWh (as presented in Fig. 9b), with the lowest from hybrid CSP with biomass from biowaste origin [48]. It is followed by CSP plants without hybridization and being hybridized with different fossil or renewable fuels and hysol CSP plant [47,48,[50], [51], [52],67]. The higher ranges of human toxicity come from dish Stirling CST in Ref. [57], at 5.85E+01 g1,4DCBeq per kWh and solar tower and parabolic trough CSP in Ref. [56], at 8.83E+01 and 9.02E+01 g1,4DCBeq per kWh, respectively. The highest human toxicity comes from CSP plant hybrid with biomass backup burner, which is 1.26E+02 g1,4DCBeq per kWh [59]. The average human toxicity is 3.16E+01 g1,4DCB per kWh, which is not much different from the median value, of 2.14E+01 g1,4DCB per kWh.