LCOE Analysis of Tower Concentrating Solar Power Plants Using Di ﬀ erent Molten-Salts for Thermal Energy Storage in China

: In recent years, the Chinese government has vigorously promoted the development of concentrating solar power (CSP) technology. For the commercialization of CSP technology, economically competitive costs of electricity generation is one of the major obstacles. However, studies of electricity generation cost analysis for CSP systems in China, particularly for the tower systems, are quite limited. This paper conducts an economic analysis by applying a levelized cost of electricity (LCOE) model for 100 MW tower CSP plants in ﬁve locations in China with four di ﬀ erent molten-salts for thermal energy storage (TES). The results show that it is inappropriate to build a tower CSP plant nearby Shenzhen and Shanghai. The solar salt (NaNO 3 -KNO 3 , 60-40 wt.%) has lower LCOE than the other three new molten-salts. In order to calculate the time when the grid parity would be reached, four scenarios for CSP development roadmap proposed by International Energy Agency (IEA) were considered in this study. It was found that the LCOE of tower CSP would reach the grid parity in the years of 2038–2041 in the case of no future penalties for the CO 2 emissions. This study can provide support information for the Chinese government to formulate incentive policies for the CSP industry.


Introduction
Concentrating solar power (CSP) technology is one of the most promising renewable energy-based electricity generation technologies to deal with the increasing demand of power consumption and environmental sustainability. It can generate renewable electricity from direct sunlight and produce nearly no greenhouse gas emissions. In addition, compared to photovoltaic (PV) cells, the CSP systems can deliver dispatchable electricity regardless of the weather conditions by integrating with thermal energy storage (TES) units. In recent years, the Chinese government has vigorously promoted the development of CSP technology. The construction of CSP plants in China has increased at a great rate since starting in 2013 [1]. The installed capacity was about 29.3 MWe by the end of 2017, which was an increase of 3.53% over that in 2016. Meanwhile, the global installation capacity increase rate was 2.3% [2].
For the commercialization of CSP technology, the low cost of the electricity generation is one of the major obstacles. Among different types of CSP technologies, the solar power tower technology can achieve higher operating temperature compared to the parabolic trough and linear Fresnel technologies. It can yield greater efficiency of thermal-to-electric conversion in the power block and result in lower cost for TES [1]. As a key factor to evaluate the economic feasibility of CSP systems, the electricity

LCOE Model for CSP Systems
The LCOE is one of the most frequently used characteristics to evaluate the electricity generation cost of different technologies, which calculates the cost of electricity during the lifetime and takes into account the time value of money and the risks. It is equal to the sum of all the discounted costs incurred during the lifetime of the project divided by the units of discounted energy produced over the entire lifetime [24] and can be expressed as follows:  Table 1. The numerator includes the capital cost of the system, land cost, operation and maintenance cost, and insurance cost. The denominator also shows the sum of the real electricity generated during the lifetime of the plant divided by the discount rate. The initial costs of the plant are considered as being paid up-front in this model,

LCOE Model for CSP Systems
The LCOE is one of the most frequently used characteristics to evaluate the electricity generation cost of different technologies, which calculates the cost of electricity during the lifetime and takes into account the time value of money and the risks. It is equal to the sum of all the discounted costs incurred during the lifetime of the project divided by the units of discounted energy produced over the entire lifetime [24] and can be expressed as follows: LCOE t = C t + L + N n=1 (V + I)C t (1 + r) −n N n=1 S · TF · η(1 − DR) n (1 + r) −n (1) where all the symbols in the equation are summarized in Table 1. The numerator includes the capital cost of the system, land cost, operation and maintenance cost, and insurance cost. The denominator also shows the sum of the real electricity generated during the lifetime of the plant divided by the discount rate. The initial costs of the plant are considered as being paid up-front in this model, thereafter the capital and land costs do not have to be discounted. However, the operation and maintenance and insurance costs should be discounted as they are paid annually throughout the lifetime of the plant, which can be evaluated at a constant percentage of the total cost of the system. In addition, the annual electricity production can be directly calculated based on the available solar resource in cooperation with the tracking factor, performance factor, and degradation rate.

Capital Cost of the CSP System
Several studies demonstrated that reduction in the capital costs of CSP plants can be characterized by the learning curves [25,26] and the cost of a system installed in a certain year, C t , can be described as a function of the global cumulative installed capacity expressed as follows [6]: where Q t is the global cumulative installed capacity in a certain year (GW); Q 0 is the cumulative installed capacity in the reference year (GW); LR is the learning rate (%); C 0 is the capital cost of the CSP system installed in the reference year (RMB W −1 ). In this paper, the year of 2017 was taken as the reference year and the capital cost of the LCOE present value was determined by C 0 which was calculated by System Advisor Model (SAM) of National Renewable Energy Laboratory (NREL) [23].

Cumulative Installed Capacity
The global cumulative installed capacity by the end of 2017 was reported to be 5.133 GW [2]. In order to evaluate the future cumulative installed capacity during the years of 2018-2050, four scenarios of CSP development roadmap proposed by IEA are used and listed in Table 2. According to Hernández-Moro et al. [6] and Li et al. [15], the logistic function of an S-shaped curve can describe the objectives of cumulative installed capacity for Blue Map scenario and a 3-order polynomial equation can fit the objectives for the other scenarios as follows: where t is the year and the parameters of r, M and k i (i = 1, 2, 3) are also summarized in Table 2.
It should be noticed that the values calculated by Equation (3) were only used for the years of 2018-2050 (shown in Figure 2) when estimating the future evolution of LCOE, while Q t equaling to Q 0 was used at the year of 2017 to calculate the present value of LCOE. It can be seen that the Global Outlook Advanced scenario is the most ambitious because it considers all useful policies to CSP technology, along the lines of the industry's recommendations, from around the world. The objective of cumulative installed capacity continues to grow fast and reaches 1524 GW by 2050. The Blue Map scenario has the greatest increase of the cumulative installed capacity during 2025 to 2040, but the objective has seen little growth after 2045 and reaches 630 GW by 2050 which is less than half of the Global Outlook Advanced scenario. Advanced scenario is the most ambitious because it considers all useful policies to CSP technology, along the lines of the industry's recommendations, from around the world. The objective of cumulative installed capacity continues to grow fast and reaches 1524 GW by 2050. The Blue Map scenario has the greatest increase of the cumulative installed capacity during 2025 to 2040, but the objective has seen little growth after 2045 and reaches 630 GW by 2050 which is less than half of the Global Outlook Advanced scenario.

Learning Rate
The range of learning rate for solar thermal power tower technology is 5%-20% in the literature [27][28][29][30]. In this study, we used a conservative estimation of 10% recommended by IEA [20,22].

Land Cost
Many geographic conditions need to be considered while selecting a potential location for installing a tower CSP plant in China, including solar resource, water resource, energy demand, and power grid accessibility [15,19]. Water is a major concern in deciding on the cooling mode. Adopting the dry cooling technology would increase the investment of 7%-9% and reduce the electricity output by 5% [31]. The energy demand is another crucial factor for site selection. The eastern region of China has the most population and the highest energy demand. However, the northwestern part has the highest DNI.

Learning Rate
The range of learning rate for solar thermal power tower technology is 5%-20% in the literature [27][28][29][30]. In this study, we used a conservative estimation of 10% recommended by IEA [20,22].

Land Cost
Many geographic conditions need to be considered while selecting a potential location for installing a tower CSP plant in China, including solar resource, water resource, energy demand, and power grid accessibility [15,19]. Water is a major concern in deciding on the cooling mode. Adopting the dry cooling technology would increase the investment of 7%-9% and reduce the electricity output by 5% [31]. The energy demand is another crucial factor for site selection. The eastern region of China has the most population and the highest energy demand. However, the northwestern part has the highest DNI. In this study, five locations with different DNI values were selected to assess the potential for installing tower CSP plants in China, which are Huizhou near Shenzhen, Taicang near Shanghai, Yanqing in Beijing, Linxi in Inner Mongolia, and Delingha in Qinghai (as shown in Figure 3). Four of them are located in eastern China where the electricity demand is relatively high, near cities with grid transmission, and sufficient water resources in order to analyze the economic feasibility of building tower CSP plants in this region, particularly near the most developed cities of China. Delingha was taken as a reference site where the annual DNI is near the bottom threshold of 2000 kWh m −2 year −1 for installing any form of the CSP systems to achieve a reasonable economic performance [32]. It can be observed from Figure 3 that Huizhou and Taicang have relatively low DNI among all the five locations. Since DNI is the most important energy input data for CSP systems in SAM, the specific DNI for those locations are calculated by the method based on the air mass which is introduced later. Then, the land areas required to build a tower CSP plant of 100 MW in different locations was calculated by SAM and shown in Table 3. It can be seen that the land price and unit land costs of Huizhou and Taicang were much higher than the other three locations. In this study, five locations with different DNI values were selected to assess the potential for installing tower CSP plants in China, which are Huizhou near Shenzhen, Taicang near Shanghai, Yanqing in Beijing, Linxi in Inner Mongolia, and Delingha in Qinghai (as shown in Figure 3). Four of them are located in eastern China where the electricity demand is relatively high, near cities with grid transmission, and sufficient water resources in order to analyze the economic feasibility of building tower CSP plants in this region, particularly near the most developed cities of China. Delingha was taken as a reference site where the annual DNI is near the bottom threshold of 2000 kWh m -2 year -1 for installing any form of the CSP systems to achieve a reasonable economic performance [32]. It can be observed from Figure 3 that Huizhou and Taicang have relatively low DNI among all the five locations. Since DNI is the most important energy input data for CSP systems in SAM, the specific DNI for those locations are calculated by the method based on the air mass which is introduced later. Then, the land areas required to build a tower CSP plant of 100 MW in different locations was calculated by SAM and shown in Table 3. It can be seen that the land price and unit land costs of Huizhou and Taicang were much higher than the other three locations.

Operation and Maintenance Cost
The operation and maintenance cost of CSP plants is lower than that of traditional power plants since fuel consumption is not necessary in CSP plants. This expense refers to the salary and welfare for the staff, the costs of water used for cooling and cleaning, plant operation costs, and field maintenance costs. The annual operation and maintenance cost was taken as 2% of the capital investment of the system, based on the reported values [6,33,34].

Insurance Cost
Since the CSP plants have relatively high technological risks in contrast with the conventional power plants, an insurance policy should be considered. The insurance cost includes fixed assets

Operation and Maintenance Cost
The operation and maintenance cost of CSP plants is lower than that of traditional power plants since fuel consumption is not necessary in CSP plants. This expense refers to the salary and welfare for the staff, the costs of water used for cooling and cleaning, plant operation costs, and field maintenance costs. The annual operation and maintenance cost was taken as 2% of the capital investment of the system, based on the reported values [6,33,34].

Insurance Cost
Since the CSP plants have relatively high technological risks in contrast with the conventional power plants, an insurance policy should be considered. The insurance cost includes fixed assets insurance and other insurance which can be assumed as 0.5% of the capital investment of the system [35].

Solar Resource
The DNI is strongly affected by the composition of atmosphere and the weather corresponding to a specific location. The DNI on the earth's surface between sunrise and sunset can be calculated according to the following experimentally determined equation [36]: where h is the location altitude above sea level (m) and AM is the air mass. AM is defined as [37]: where α is the solar elevation angle which can be calculated as a function of the solar declination angle, δ, local latitude, φ, and solar hour angle, γ, [38]: The DNIs of summer solstice in Huizhou, Taicang, Yanqing, Linxi, and Delingha were calculated by the above mentioned method and shown in Figure 4. It can be seen that the DNI at Linxi is higher than that at the other locations except for the reference site Delingha. insurance and other insurance which can be assumed as 0.5% of the capital investment of the system [35].

Solar Resource
The DNI is strongly affected by the composition of atmosphere and the weather corresponding to a specific location. The DNI on the earth's surface between sunrise and sunset can be calculated according to the following experimentally determined equation [36]: where h is the location altitude above sea level (m) and AM is the air mass. AM is defined as [37]: where α is the solar elevation angle which can be calculated as a function of the solar declination angle, δ, local latitude, ϕ, and solar hour angle, γ, [38]: ( ) 1 sin sin sin cos cos cos The DNIs of summer solstice in Huizhou, Taicang, Yanqing, Linxi, and Delingha were calculated by the above mentioned method and shown in Figure 4. It can be seen that the DNI at Linxi is higher than that at the other locations except for the reference site Delingha. Then, the annual DNI, S, as the integral of DNI along a year can be calculated as follows: where t1 and t2 indicate the beginning and end of the year, respectively. The annual DNI of Huizhou, Taicang, Yanqing, Linxi, and Delingha are shown in Table 4. It indicates that the annual DNI in Delingha is much higher than other sites. The annual DNIs in Taicang and Huizhou are even lower than 800 kWh m -2 year -1 .  Then, the annual DNI, S, as the integral of DNI along a year can be calculated as follows: where t 1 and t 2 indicate the beginning and end of the year, respectively. The annual DNI of Huizhou, Taicang, Yanqing, Linxi, and Delingha are shown in Table 4. It indicates that the annual DNI in Delingha is much higher than other sites. The annual DNIs in Taicang and Huizhou are even lower than 800 kWh m −2 year −1 .

Tracking Factor
As the CSP systems can only use DNI, the tracking system is critical for CSP plants. For tower systems, azimuth-altitude dual axis solar tracking mechanisms are employed for the heliostat. Hence, the tracking accuracy to the sunlight is high and the tracking factor is one [33].

Performance Factor
The performance factor can be calculated as follows: For CSP systems, the utilized solar resource approximates to the DNI. And the electricity production per installed watt can be obtained by dividing the annual electrical output to the power of the plant. The electrical output was estimated by SAM, which was not characterized by the solar field but by the power of the turbines. In addition, since value of the performance factor is relevant to TES, the range of 0-10 h of TES capacity was evaluated.

Degradation Rate
Because of degradation, the annual electrical output of CSP systems would be slightly diminished. The recommended value of degradation rate was 0.2% mainly owing to the degradation of turbines [7].

Discount Rate
The discount rate is used to consider the time value of money and the risk of the investment. According to IEA [34], the discount rate is between 10% and 15% for CSP systems. A conservative value of 10% was adopted in the present paper.

Lifetime of the System
For CSP systems, the common lifetime is in the range of 25-40 years [22,33,34,39,40]. In this study, a lifetime of 30 years was used as suggested by most similar studies.

Results and Discussion
On the basis of the LCOE model described above, the economic analysis of 100 MW tower CSP plants in five locations in China with four different molten-salts for TES was estimated. Calculation results of both the LCOE present value in 2017 and its future evolution over the period of 2018-2050 under four scenarios proposed by the IEA were also presented. Moreover, the specific time when CSP grid parities can be attained was estimated and influences of the land cost, TES capacity, learning rate, discount rate, lifetime, degradation rate, operation-maintenance cost, and insurance cost on the LCOE were examined. As the reference, a 100 MW plant using solar salt (NaNO 3 -KNO 3 , 60-40 wt.%) with 6 h TES in Delingha was used in this study. Unless otherwise stated, the values for the input parameters of Equation (1) were used as specified in Table 1. Table 5 shows the initial costs and LCOE present values for different locations under reference conditions. It can be seen that the initial costs of Huizhou and Taicang were much higher than that of Delingha which is mainly due to expensive land cost. The influence of land cost on the LCOE is illustrated in Figure 5. The x axis shows the percentage of the land cost reduction. It was found that the land cost affected the LCOE significantly for Huizhou and Taicang as their land costs accounted Energies 2019, 12, 1394 9 of 17 for 38.8% and 30.8% of their total initial capital investment, respectively. Furthermore, the LCOE in Huizhou and Taicang was still higher than 6 RMB kWh −1 even if the land cost dropped to zero. It was mainly because the DNI in both locations is less than 800 kWh m −2 year −1 , which makes the capacity factor relatively low. Thus, it is inappropriate to install a tower CSP plant nearby Shenzhen and Shanghai regardless of the land cost. It can also be observed from Figure 5 that the land cost had a slight influence on the LCOE of CSP plants in Yanqing, Linxi, and Delingha. Yanqing is the selected location of the first pilot CSP plant in Beijing and the land cost is quite low. The LCOE of Linxi had the lowest value of 2.33 RMB kWh −1 because of its highest DNI among the four selected locations at the eastern of China. Therefore, it can be concluded that the major factor resulting in low LCOE for different locations is the high DNI, which leads to more produced energy, higher capacity factor, and more efficient TES. the eastern of China. Therefore, it can be concluded that the major factor resulting in low LCOE for different locations is the high DNI, which leads to more produced energy, higher capacity factor, and more efficient TES.  The TES unit is an important sub-system of CSP plants which guarantees dispatchable electricity supply. Moreover, high heat transfer fluid (HTF) operating temperature in a CSP plant is beneficial because it can improve the heat to electricity conversion efficiency. The U.S. Department of Energy (DOE) suggested that the appropriate HTF candidates must be thermally stable at temperatures up to 720 °C [41]. In the present study, four molten-salt eutectic mixtures were selected as the energy storage media to estimate the corresponding LCOE. Except for solar salt which was selected as the reference, the other three salts all have the temperature of thermal stability higher than 715 °C and show the potential to be used in commercial CSP plants which has been reported in previous studies [42][43][44][45][46]. Their thermophysical properties and TES costs are listed in Table 6 [1, [42][43][44][45][46][47]. It can be seen that the chloride-salt mixture (NaCl-KCl-ZnCl2, 8.1-31.3-60.6 wt.%) has the lowest TES cost due to the lowest salt price and largest density. And the TES costs of the fluoride-salt mixture (LiF-NaF-KF, 29.3-11.7-59 wt.%) and carbonate-salt mixture (Li2CO3-Na2CO3-K2CO3, 32.1-33.4-34.5 wt.%) are much higher since their salt prices are more than three times of the prices of solar salt and chloride-salt mixture. Table 6. Thermal properties and thermal energy storage (TES) costs of selected molten-salts.

Molten-Salt
NaNO3-KNO3 LiF-NaF-KF Li2CO3-Na2CO3-K2CO3 NaCl-KCl-ZnCl2 29.3-11.7- The TES unit is an important sub-system of CSP plants which guarantees dispatchable electricity supply. Moreover, high heat transfer fluid (HTF) operating temperature in a CSP plant is beneficial because it can improve the heat to electricity conversion efficiency. The U.S. Department of Energy (DOE) suggested that the appropriate HTF candidates must be thermally stable at temperatures up to 720 • C [41]. In the present study, four molten-salt eutectic mixtures were selected as the energy storage media to estimate the corresponding LCOE. Except for solar salt which was selected as the reference, the other three salts all have the temperature of thermal stability higher than 715 • C and show the potential to be used in commercial CSP plants which has been reported in previous studies [42][43][44][45][46]. Their thermophysical properties and TES costs are listed in Table 6 [1, [42][43][44][45][46][47]. It can be seen that the chloride-salt mixture (NaCl-KCl-ZnCl 2 , 8.1-31.3-60.6 wt.%) has the lowest TES cost due to the lowest salt price and largest density. And the TES costs of the fluoride-salt mixture (LiF-NaF-KF, 29.3-11.7-59 wt.%) and carbonate-salt mixture (Li 2 CO 3 -Na 2 CO 3 -K 2 CO 3 , 32.1-33.4-34.5 wt.%) are much higher since their salt prices are more than three times of the prices of solar salt and chloride-salt mixture. Table 6. Thermal properties and thermal energy storage (TES) costs of selected molten-salts.  Figures 6 and 7 indicate the effect of TES capacity on the performance factor and LCOE present value using different salt mixtures as the energy storage media under reference conditions. It can be found that the performance factor of solar salt which has a much lower temperature of thermal stability is higher than that of the fluoride-salt mixture and chloride-salt mixture. The chloride-salt mixture with the lowest performance factor has the highest LCOE when the CSP system has no TES. Since a larger capacity TES unit improves the performance factor of CSP systems as confirmed by other studies [48,49] then the LCOE of all molten-salts decreases as the capacity of TES increases. However, when the performance factor no longer improves along with the increasing TES capacity, then LCOE goes up, thereafter, as the cost of the TES rises. Thus, there is an optimal capacity of TES resulting in the lowest LCOE. The LCOE of the chloride-salt mixture reduces sharply before the TES capacity reaches the optimal value because of its comparably lowest salt price. As such, the increasing trends of the LCOEs of the fluoride-salt and carbonate-salt mixtures are more remarkable when the TES capacity is higher than the optimal value due to their high costs. The solar salt has the lowest LCOE among those four molten-salts regardless of the TES capacity.  Figure 6 and Figure 7 indicate the effect of TES capacity on the performance factor and LCOE present value using different salt mixtures as the energy storage media under reference conditions. It can be found that the performance factor of solar salt which has a much lower temperature of thermal stability is higher than that of the fluoride-salt mixture and chloride-salt mixture. The chloride-salt mixture with the lowest performance factor has the highest LCOE when the CSP system has no TES. Since a larger capacity TES unit improves the performance factor of CSP systems as confirmed by other studies [48,49] then the LCOE of all molten-salts decreases as the capacity of TES increases. However, when the performance factor no longer improves along with the increasing TES capacity, then LCOE goes up, thereafter, as the cost of the TES rises. Thus, there is an optimal capacity of TES resulting in the lowest LCOE. The LCOE of the chloride-salt mixture reduces sharply before the TES capacity reaches the optimal value because of its comparably lowest salt price. As such, the increasing trends of the LCOEs of the fluoride-salt and carbonate-salt mixtures are more remarkable when the TES capacity is higher than the optimal value due to their high costs. The solar salt has the lowest LCOE among those four molten-salts regardless of the TES capacity. Figure 6. Influence of TES capacity on the performance factor for four different molten-salts. Figure 6. Influence of TES capacity on the performance factor for four different molten-salts.    8 also illustrates the time when the electricity cost of tower CSP plants will reach the grid parity under reference conditions. Since more than three quarters of the electricity is generated by coal-fired power plants in China [34], the cost evolution of coal-fired electricity was used as the grid parity in this study. The cost of electricity produced by coal-fired power plants is 0.41 RMB kWh -1 [50] which is only considered the production costs and would increase linearly in the future with an annual conservative growth rate of 2.7% [15]. In addition, considering the CO2 emissions by the coalfired power plants, the carbon emission prices of 162.5 RMB ton -1 CO2 [51] and 325 RMB ton -1 CO2 [20] were added to estimate the grid parities with an emission factor of 0.9 kg CO2 kWh -1 [34,52,53].      8 also illustrates the time when the electricity cost of tower CSP plants will reach the grid parity under reference conditions. Since more than three quarters of the electricity is generated by coal-fired power plants in China [34], the cost evolution of coal-fired electricity was used as the grid parity in this study. The cost of electricity produced by coal-fired power plants is 0.41 RMB kWh -1 [50] which is only considered the production costs and would increase linearly in the future with an annual conservative growth rate of 2.7% [15]. In addition, considering the CO2 emissions by the coalfired power plants, the carbon emission prices of 162.5 RMB ton -1 CO2 [51] and 325 RMB ton -1 CO2 [20] were added to estimate the grid parities with an emission factor of 0.9 kg CO2 kWh -1 [34,52,53].  8 also illustrates the time when the electricity cost of tower CSP plants will reach the grid parity under reference conditions. Since more than three quarters of the electricity is generated by coal-fired power plants in China [34], the cost evolution of coal-fired electricity was used as the grid parity in this study. The cost of electricity produced by coal-fired power plants is 0.41 RMB kWh −1 [50] which is only considered the production costs and would increase linearly in the future with an annual conservative growth rate of 2.7% [15]. In addition, considering the CO 2 emissions by the coal-fired power plants, the carbon emission prices of 162.5 RMB ton −1 CO 2 [51] and 325 RMB ton −1 CO 2 [20] were added to estimate the grid parities with an emission factor of 0.9 kg CO 2 kWh −1 [34,52,53]. According to Figure 8, the time when the grid parities of tower CSP systems would be achieved are  Table 7. It can be observed that the LCOE of tower CSP would reach the grid parity in the years of 2038-2041 in the case of no future penalties for CO 2 emissions. And the grid parity would bring forward about 7-8 years for the carbon emission price of 162.5 RMB ton −1 CO 2 and about 13-14 years for the carbon emission price of 325 RMB ton −1 CO 2 . In order to evaluate the influence of learning rate on the LCOE for tower CSP systems, the LCOE in the year of 2050 with the learning rate ranging from 5% to 20% under reference conditions are estimated at four scenarios and shown in Figure 9. It illustrates that the LCOE reduces as the learning rate increases and the reduction rates are 66.4%, 71.8%, 68.2%, and 69.9% for Blue map scenario, Global outlook advanced scenario, Global outlook moderate scenario, and Roadmap scenario, respectively. The result indicates that this parameter has a significant effect on the economic feasibility of tower CSP plants. According to Figure 8, the time when the grid parities of tower CSP systems would be achieved are summarized in Table 7. It can be observed that the LCOE of tower CSP would reach the grid parity in the years of 2038-2041 in the case of no future penalties for CO2 emissions. And the grid parity would bring forward about 7-8 years for the carbon emission price of 162.5 RMB ton -1 CO2 and about 13-14 years for the carbon emission price of 325 RMB ton -1 CO2. In order to evaluate the influence of learning rate on the LCOE for tower CSP systems, the LCOE in the year of 2050 with the learning rate ranging from 5% to 20% under reference conditions are estimated at four scenarios and shown in Figure 9. It illustrates that the LCOE reduces as the learning rate increases and the reduction rates are 66.4%, 71.8%, 68.2%, and 69.9% for Blue map scenario, Global outlook advanced scenario, Global outlook moderate scenario, and Roadmap scenario, respectively. The result indicates that this parameter has a significant effect on the economic feasibility of tower CSP plants. Some studies in the literature [49,54,55] claimed that the discount rate is one of the most significant variables affecting the LCOE of the CSP systems, which reflects the risk-adjusted opportunity of capital investment. Figure 10 shows that the present value of LCOE under reference conditions rises about 35% when the discount rate increases from 10% to 15%. Although the discount rate for the CSP technology is relatively high for now, it is expected to decline in the future as CSP markets develop and the related technologies become more mature. Some studies in the literature [49,54,55] claimed that the discount rate is one of the most significant variables affecting the LCOE of the CSP systems, which reflects the risk-adjusted opportunity of capital investment. Figure 10 shows that the present value of LCOE under reference conditions rises about 35% when the discount rate increases from 10% to 15%. Although the discount rate for the CSP technology is relatively high for now, it is expected to decline in the future as CSP markets develop and the related technologies become more mature. In order to study the sensitivity of the LCOE with the lifetime of tower CSP plants, the LCOE present value in different lifetime expectations under reference conditions were estimated and shown in Figure 11. It can be seen that the LCOE is less sensitive to the lifetime. As the life time increases from 25 to 40 years, the LCOE only reduces by 5.6%. Figure 11. Effect of lifetime on the LCOE. Figure 12 displays the impacts of degradation rate, operation-maintenance cost, and insurance cost on the LCOE for tower CSP systems. The variables are changed within the range of -20% to 20% with an interval of 5% around reference conditions. It can be seen that the effect of operationmaintenance cost is greater than that of other two factors and the influence of the degradation rate is the smallest. One way to reduce the operation-maintenance cost is to improve the level of management and use high quality CSP equipment. In order to study the sensitivity of the LCOE with the lifetime of tower CSP plants, the LCOE present value in different lifetime expectations under reference conditions were estimated and shown in Figure 11. It can be seen that the LCOE is less sensitive to the lifetime. As the life time increases from 25 to 40 years, the LCOE only reduces by 5.6%. In order to study the sensitivity of the LCOE with the lifetime of tower CSP plants, the LCOE present value in different lifetime expectations under reference conditions were estimated and shown in Figure 11. It can be seen that the LCOE is less sensitive to the lifetime. As the life time increases from 25 to 40 years, the LCOE only reduces by 5.6%. Figure 11. Effect of lifetime on the LCOE. Figure 12 displays the impacts of degradation rate, operation-maintenance cost, and insurance cost on the LCOE for tower CSP systems. The variables are changed within the range of -20% to 20% with an interval of 5% around reference conditions. It can be seen that the effect of operationmaintenance cost is greater than that of other two factors and the influence of the degradation rate is the smallest. One way to reduce the operation-maintenance cost is to improve the level of management and use high quality CSP equipment.  Figure 12 displays the impacts of degradation rate, operation-maintenance cost, and insurance cost on the LCOE for tower CSP systems. The variables are changed within the range of −20% to 20% with an interval of 5% around reference conditions. It can be seen that the effect of operation-maintenance cost is greater than that of other two factors and the influence of the degradation rate is the smallest. One way to reduce the operation-maintenance cost is to improve the level of management and use high quality CSP equipment. Energies 2019, 12, x FOR PEER REVIEW 14 of 17 Figure 12. Sensitivity analysis of degradation rate, operation-maintenance cost, and insurance cost.

Conclusions
In order to provide support information for the Chinese government to formulate incentive policies for the CSP industry, we have presented the economic analysis based on the LCOE model (1) It is inappropriate to build a tower CSP plant nearby Shenzhen and Shanghai. Even if the land costs drop to zero, the LCOEs would be still higher than 6 RMB kWh -1 mainly owing to the relatively low DNI of less than 800 kWh m -2 year -1 . The impact of DNI on the LCOE is much more significant than the other influence factors. (2) There is an optimal capacity of TES resulting in the lowest LCOE for a certain tower CSP plant, as the performance factor increases with the increasing TES capacity and then becomes stable. The solar salt has the lowest LCOE among those four molten-salts regardless of the TES capacity. (3) The LCOE of tower CSP would reach the grid parity in the years of 2038-2041 in the case of no future penalties for the CO2 emissions based on the four scenarios for CSP development roadmap proposed by IEA. As such, the grid parity would be brought forward about 7-8 years for the carbon emission price of 162.5 RMB ton -1 CO2 and about 13-14 years for the carbon emission price of 325 RMB ton -1 CO2. (4) The LCOE of the tower CSP systems is significantly affected by the learning rate and discount rate, yet it is less sensitive to the lifetime. Moreover, the impact of operation-maintenance cost is greater than that of insurance cost and degradation rate.
Incentive policy is a crucial force for the rapid development of the CSP industry and reducing the LCOE of CSP plants. Further studies can be conducted to evaluate the impacts of incentive policies for CSP projects in China, such as investment and production taxes credit, cash incentives, sales, or use tax reduction and different depreciation modes.

Conclusions
In order to provide support information for the Chinese government to formulate incentive policies for the CSP industry, we have presented the economic analysis based on the LCOE model for 100 MW tower CSP plants in the locations of Huizhou near Shenzhen, Taicang near Shanghai, Yanqing in Beijing, Linxi in Inner Mongolia, and Delingha in Qinghai in China with four molten-salt eutectic mixtures of the chloride-salt mixture (NaCl-KCl-ZnCl 2 , 8.1-31.3-60.6 wt.%), fluoride-salt mixture (LiF-NaF-KF, 29.3-11.7-59 wt.%), carbonate-salt mixture (Li 2 CO 3 -Na 2 CO 3 -K 2 CO 3 , 32.1-33.4-34.5 wt.%), and solar salt (NaNO 3 -KNO 3 , 60-40 wt.%) as the energy storage media. Major conclusions were summarized as follows: (1) It is inappropriate to build a tower CSP plant nearby Shenzhen and Shanghai. Even if the land costs drop to zero, the LCOEs would be still higher than 6 RMB kWh −1 mainly owing to the relatively low DNI of less than 800 kWh m −2 year −1 . The impact of DNI on the LCOE is much more significant than the other influence factors. (2) There is an optimal capacity of TES resulting in the lowest LCOE for a certain tower CSP plant, as the performance factor increases with the increasing TES capacity and then becomes stable. The solar salt has the lowest LCOE among those four molten-salts regardless of the TES capacity. (3) The LCOE of tower CSP would reach the grid parity in the years of 2038-2041 in the case of no future penalties for the CO 2 emissions based on the four scenarios for CSP development roadmap proposed by IEA. As such, the grid parity would be brought forward about 7-8 years for the carbon emission price of 162.5 RMB ton −1 CO 2 and about 13-14 years for the carbon emission price of 325 RMB ton −1 CO 2 . (4) The LCOE of the tower CSP systems is significantly affected by the learning rate and discount rate, yet it is less sensitive to the lifetime. Moreover, the impact of operation-maintenance cost is greater than that of insurance cost and degradation rate.
Incentive policy is a crucial force for the rapid development of the CSP industry and reducing the LCOE of CSP plants. Further studies can be conducted to evaluate the impacts of incentive policies for CSP projects in China, such as investment and production taxes credit, cash incentives, sales, or use tax reduction and different depreciation modes.