Energy Conversion and Management 308 (2024) 118389

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Theoretical and run-test investigation on a 50 MW beam-down

concentrating solar power plant in China
Xinzhuang Gu a, b, Hao Chen c, Dequan Xu a, b, Shixiong Song c, Wentao Xie c, *, Yuda Chen c,
Raúl Navío Gilaberte d, Teng Jia a, b, e, *, Yanjun Dai a, b, *, Bo Yu f, Shuochen Zhou g
a
Institute of Refrigeration and Cryogenics, Shanghai Jiao Tong University, Shanghai 200240, China
b
Engineering Research Center of Solar Energy and Refrigeration, MOE, China
c
Xinchen (Shanghai) CSP New Energy Co., Ltd, Shanghai, China
d
Alia Energy Consulting S.L., Madrid, Spain
e
Department of Chemical Engineering, Imperial College London, South Kensington Campus, London SW7 2AZ, UK
f
Southwest Electric Power Design Institute Co., Ltd of China Power Engineering Consulting Group, Chengdu 610021, China
g
Shanghai Investigation, Design & Research Institute Co., Ltd, Shanghai 200434, China

A R T I C L E I N F O A B S T R A C T

Keywords: The beam-down concentrating solar power plant has the advantages of high concentrating ratio, low installation
Beam-down power plant and maintenance requirements, and low pump consumption. To conduct further theoretical and run-test eval­
Heliostat efficiency uations of the optical and thermal performance of the beam-down concentrating solar power system, the Yumen
Average flux density
Xinneng 50 MW beam-down power plant is analyzed in this research. The variation of the heliostat field effi­
Power generation
ciency, hyperboloid mirror flux density, receiver flux density, and annual power generation under various
operating parameters is discussed. The run-test results from June 12th indicated an average outlet temperature of
molten salt reached 559◦ C between 13:07 and 13:56 under an average direct normal irradiation of 739.7 W/m2.
The corresponding inlet temperature increased from 302.3◦ C to 307.6◦ C due to the rewarming effect caused by
the increase in hot molten salt. Regarding the simulation results, the recommended design parameters for the
vertex ratio of the hyperboloid mirror are between 0.775 and 0.8 to mitigate the contradiction between initial
investment and optical efficiency. The maximum values of cosine efficiency, shading and blocking efficiency,
shading efficiency of the hyperboloid mirror, and attenuation efficiency for the heliostat field are 85.56%,
99.96%, 99.35%, and 97.46% on June 1st for the maximum solar elevation angle of 70.72◦ . Consequently, the
maximum values of the average hyperboloid mirror flux density and receiver flux density are 11.2 kW/m2 and
399.1 kW/m2 on June 1st. The maximum power generation for a 7-day period is 4674.22 MWe in June with a
duration time of 100 hours. This paper can further expand research on the comprehensive performance of beam-
down power plants.

have gained significant attention in the field of solar thermal power

1. Introduction (STP). Moreover, the utilization of TES for various CSP systems is
contributed to operating in adverse weather conditions, particularly
Recently, the application of renewable energy for power generation during nighttime and cloudy days. Consequently, it is crucial to evaluate
has promoted the implementation of the dual carbon policy, which al­ the operational characteristics of CSP systems for further enhancing the
leviates the energy challenges associated with fossil fuels. As the main flexibility of the power grid.
environmental-friendly power generation technologies, wind and solar As described in Fig. 1(a), the STP technologies are categorized as: (a)
power are projected to contribute 33% and 39% of China’s electricity parabolic dish system, (b) parabolic trough system, (c) linear fresnel
generation by 2050 [1]. However, wind and photovoltaic power gen­ reflector system, (d) central receiver system, and (e) beam-down CSP
eration may have impacts on the power grid due to the intermittent system. The parabolic trough system has gained significant commercial
nature. To improve the stability of the power grid, various concentrating application at the cost of low operational temperature (<400◦ C),
solar power (CSP) systems integrated with thermal energy storage (TES) hampering the efficiency of the power cycle [2]. Similarly, the linear

* Corresponding authors at: Institute of Refrigeration and Cryogenics, Shanghai Jiao Tong University, Shanghai 200240, China (T. Jia and Y. Dai).
E-mail addresses: [email protected] (W. Xie), [email protected] (T. Jia), [email protected] (Y. Dai).

https://doi.org/10.1016/j.enconman.2024.118389
Received 7 January 2024; Received in revised form 30 March 2024; Accepted 1 April 2024
Available online 13 April 2024
0196-8904/© 2024 Elsevier Ltd. All rights reserved.
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Nomenclature Subcript
1–4 state points in extraction
asaa solar azimuth angle, ◦ at attenuation efficiency
asea solar elevation angle, ◦ c cleanliness
chm semi-minor axis, m cos cosine efficiency
Cp specific heat, J/(kg•K) e extraction
d distance, m ev evaporator
e eccentricity f1 upper focal point
f vertex ratio f2 lower focal point
h enthalpy, J/kg h heliostat
H height, m hf heliostat field
hc heat transfer coefficient, W/(m2•K) hm hyperboloid mirror
Iav average flux density (kW/m2) i inlet
L length, m in intercept efficiency
m mass flow rate, kg/s max maximum
n number min minimum
Q heat flow, W ms molten salt
R radius, m o outlet
s area, m2 pc power cycle
T temperature, oC r reflectivity
u specific internal energy, J/kg re receiver
V volume, m3 rh reheater
W output work, W sb shading and blocking efficiency
sh superheater
Greek symbols sm steam
η efficiency
θ angle between two vectors, rad Abbreviations
γ acceptance angle, rad CPC compound parabolic concentrator
φ zenith angle, rad CSP concentrating solar power
σ Stefan-Boltzmann constant, W/(m2•K4) DNI direct normal irradiation
ε emissivity HPT high-pressure turbine
ρ density of air, kg/m3 LPT low-pressure turbine
ϕ latitude STP solar thermal power
δ declination angle, rad TES thermal energy storage
ω solar hour angle, rad

fresnel reflector system has the lowest cost and lowest annual efficiency corresponding aperture radius was 7 m at the heliostat radius of 5 m.
ranging from 9% to 11% [3]. The parabolic dish system offers the Saldivia et al. [8] simulated the flux distribution of the hyperboloid
advantage of the highest concentration ratio of 1000 to 3000 and a mirror and receiver to propose an optimized CPC structure. The opti­
power cycle efficiency of 30% to 40% coupled with a stirling engine, but mized ratio of focus and vertex for the hyperboloid mirror was 0.83, and
the lowest installed capacity of 0.01–0.4 MW due to structural limita­ the recommended concentration ratio for the 3-hexagon CPC was 2. The
tions [3]. The central receiver system boasts a maximum installed ca­ simulation results demonstrated optical efficiency and power generation
pacity ratio of 63.1% [4], with the advantage of a high operating efficiency of 57% and 20.3%, respectively. Wei et al. [9] performed ray
temperature and concentration ratio. However, the tower of the central tracing simulations on a beam-down CSP system using an improved
receiver system is relatively high, resulting in higher requirements for HFLD algorithm, which included 31 heliostats, a hyperboloid mirror,
installation and maintenance as well as higher pump energy consump­ and a CPC. The simulation results were basically consistent with those
tion. For the purpose of overcoming the above shortcomings, Rabl [5] obtained using the Zemax software. The optimized parameters of aper­
first proposed the beam-down CSP system in 1976 by adopting a hy­ ture, length, and collecting angle for the CPC were determined to be
perboloid mirror to reflect concentrated sunlight onto a ground-based 1.036 m, 2.086 m, and 18◦ , respectively. In summary, it is crucial to
receiver. In this respect, several researchers have conducted the per­ conduct further theoretical and run-test evaluations of the optical per­
formance analyses of the beam-down CSP system. Segal and Epstein [6] formance of the beam-down CSP system (e.g., heliostat, hyperboloid
comparatively analyzed the effect of ellipsoid mirror and hyperboloid mirror, and receiver) to provide a comprehensive approach for the
mirror on the optical performance for the beam-down CSP system. The commercial power plant.
reason for the recommending the hyperboloid mirror is that the area of However, many research groups have focused on the beam-down
the ellipsoid mirror is 1.5 times that of the hyperboloid mirror at a CSP plants with low installed capacity (0.1 MW to 2 MW) due to the
constant ratio of f2 and f1. The installation height of the ellipsoid mirror challenges associated with accurately controlling multiple heliostats
is 2 times that of the hyperboloid mirror, resulting in more optical er­ (over 2,000 sets), reducing the cost of the secondary reflection tower
rors. Leonardi [7] evaluated the influence of eccentricity on the hyper­ (over 50 m), and designing the receiver thermal capacity (over 10 MWt),
boloid radius of the beam-down solar power plant. The simulation resulting in lower power generation. A summary of the characteristics of
results showed that the area of the hyperboloid mirror at an eccentricity demonstration and commercial beam-down CSP plants is provided in
of 3 was 4.2 times that at an eccentricity of 1.5. The recommended Fig. 1(b) and Table 1, with plants located in China, Italy, Israel, Japan,
values of acceptance angle and truncation angle for the compound and the United Arab Emirates. The thermal design power of these plants
parabolic concentrator (CPC) were 38.56◦ and 107.69◦ . The varies in the range of 0.1 MW to 50 MW, with the corresponding number

2
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

of heliostats ranging from 33 to 39,045. Grange et al. [10] reported on a Furthermore, the accuracy of the mathematical model is validated by
100 kW beam-down CSP system built in 2009 at Masdar Institute. The comparing the heliostat field efficiency and flux density with existing
experimental results indicated that the flux density of the receiver references. The run-test operation of the heliostat field, hyperboloid
aperture was up to 600 suns, and the corresponding thermal loss was mirror, and receiver on June 12th is detailed in Section 4. Subsequently,
5.5% of the thermal capacity. However, the maximum operating tem­ the comprehensive performance of the entire beam-down CSP plant is
perature of this system was limited to 393◦ C to avoid oil degradation. simulated and discussed in Section 5. Finally, the key conclusions drawn
Kodama et al. [11] designed a novel 100 kW beam-down system for from this study are listed in Section 6.
hydrogen production from splitting water at the University of Miyazaki
in 2012. The maximum power was 113.2 kW at the direct nominal ra­ 2. Fundamentals of the beam-down power plant
diation of 894.7 W/m2. The corresponding operating temperature and
peak flux density were 1,400◦ C and 495.2 kW/m2, respectively. How­ As described in Section 2, the basic information of the Yumen Xin­
ever, the secondary reflector type was an elliptic mirror, resulting in a neng 50 MW beam-down CSP plant is depicted in Section 2.1, including
tower height of 14 m. Segal [12] developed a 650 kW beam-down CSP the geographical location, photograph, and designed parameters of the
system combined with 64 heliostats and a hyperboloid mirror area of 75 entire power plant. The schematic diagram and operating principle of
m2, which was built in 2000 at the Weizmann Institute of Science. The key components in the entire power plant are presented in Section 2.2.
entrance radius, height, and magnification factor of the CPC were 1.1 m,
5 m, and 25 respectively, which enhanced the average concentration of
2.1. Basic information
the receiver to 4000. Magaldi Co., Ltd [13] demonstrated a 2 MW beam-
down CSP system built in Messina in 2016, which consisted of 786 he­
As illustrated in Fig. 2, the Yumen Xinneng 50 MW beam-down CSP
liostats. The experimental results revealed that 270 tons of sand could be
plant is located in Yumen city, Gansu province, which spans a latitude
heated to 550–650◦ C for storing 8.2 MWh of thermal energy. Moreover,
range of 32◦ 31′ N to 42◦ 57′ N and a longitude range of 92◦ 13′ E to
Xinchen CSP Co., Ltd constructed the Yumen Xinneng 50 MW and
108◦ 46′ E. The total solar radiation of Gansu Province varies from 4,700
Yancheng 300 kW beam-down power plants in 2017 and 2015,
MJ/m2 to 6,350 MJ/m2, in which Yumen city is situated in a region
respectively. The heliostat numbers for the two plants were 39,045 and
boasting the most abundant solar energy resources (over 6,300 MJ/m2)
160, with corresponding tower heights of 64 m and 20 m. However,
in China [14]. The climate type in Yumen city is classified as semi-arid
there have been no reports on the theoretical and run-test evaluation of a
climate in the middle temperature zone. The characters of the semi-arid
50 MW commercial beam-down CSP plant in open literature. Therefore,
climate are low precipitation, dry air, and abundant sunshine, which are
exhaustive research on a 50 MW beam-down CSP plant is needed to
conducive to the utilization of solar energy. The aerial view of the
provide a comprehensive approach for commercial power plant
Yumen Xinneng 50 MW beam-down CSP plant is depicted in Fig. 2(c),
applications.
situated at a latitude of 40◦ 33′ N and a longitude of 97◦ 27′ E. The cor­
Theoretical and run-test evaluations of a 50 MW beam-down CSP
responding power station land area is 3.69 square kilometers.
plant are carried out in this paper to describe the operating performance
The annual design power generation of the Yumen Xinneng 50 MW
of heliostat, hyperboloid mirror, receiver, and power cycle. In this
beam-down CSP power plant is 215 million kW•h to meet the electricity
respect, the configuration of components and schematic diagram of the
demand of 20,000 households. As listed in Fig. 3, the entire power plant
Yumen Xinneng 50 MW beam-down CSP plant are presented in Section
consists of several components: (a) molten salt tank, (b) steam turbine,
2. The calculation equations for key components and the theoretical
(c) heliostat field, (d) hyperboloid mirror, (e) receiver, and (f) molten
solution procedures used in the model are described in Section 3.
salt pipes. With respect to the molten salt tank, the working fluid type

Fig. 1. Characteristics of solar thermal power technologies and beam-down CSP plants.

3
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Table 1
Summarization literatures on the solar thermal power system.
Authors Solar thermal power type Method Capacity Concentrating ratio Temperature Reference

M. Romero and A. Steinfeld Parabolic trough system Numerical 10–200 MW 70–80 390 C
◦
[2]
H. Müller-Steinhagen and F. Trieb Linear fresnel reflector system Numerical 10–200 MW 25–100 390◦ C [3]
H. Müller-Steinhagen and F. Trieb Parabolic dish system Numerical 0.01–0.4 MW 1000–3000 750◦ C [3]
Z. Wang Central receiver system Numerical 10–150 MW 300–1000 565◦ C [4]
A. Rabl Beam-down CSP system Numerical [-] 2500 1300 K [5]
A. Segal and M. Epstein Beam-down CSP system Numerical 6.4 MW and 44.5 MWa 270–3020 [-] [6]
E. Leonardi Beam-down CSP system Numerical 26.8569 MWha 0–1000 [-] [7]
Saldivia et al. Beam-down CSP system Numerical 10 MW 2000 800 K-1000 K [8]
Wei et al. Beam-down CSP system Numerical [-] 0–4465 [-] [9]
Grange et al. Beam-down CSP system Experimental 0.1 MW 150–600 250◦ C-550◦ C [10]
Kodama et al. Beam-down CSP system Experimental 0.113 MW 1400 495.2 [11]
Segal Beam-down CSP system Experimental 0.65 MW 4000 [-] [12]
Magaldi Co., Ltd Beam-down CSP system Experimental 2 MW [-] 550◦ C-650◦ C [13]
Gu et al. Beam-down CSP system Experimental 50 MW 0–1500 565◦ C This research

a. the receiver capacity.

Fig. 2. (a) The distribution map of global horizontal irradiation in China [14]; (b) the geographical location of Gansu province [16]; (c) the photograph of the Yumen
Xinneng 50 MW beam-down CSP plant.

and heat storage time are solar salt (40% KNO3 + 60 %NaNO3) and 9 h, Fig. 4, comprising five distinct subsystems: (a) heliostat field, (b) hy­
respectively. The annual design utilization hours and cycle efficiency of perboloid mirror, (c) receiver, (d) molten salt tank, and (e) power cycle.
the steam turbine are 4,300 h and 43.7% [15], respectively. The helio­ The heliostat field focuses sunlight onto the upper focal point of the
stat field comprises 2,603 heliostats divided into the north, west, and hyperboloid mirror, which subsequently reflects the concentrated sun­
east regions. Each heliostat has dimensions of 4.4 m in length, 4 m in light onto the entrance surface of the receiver at the lower focal point.
width, and 2.2 m in height. For the hyperboloid mirror, the lowest The concentrated sunlight raises the temperature of the receiver to over
height, highest height, and maximum radius are 60 m, 74 m, and 34 m, 300◦ C, effectively preventing the solidification of the molten salt. The
respectively. With respect to the receiver, the design thermal capacity operating strategy of the molten salt tank subsystem prioritizes the
and height are 17.1 MWt and 11 m, respectively. The detailed param­ storage of the heated working fluid prior to entering into the power cycle
eters of the above key equipment are summarized in Table 2. subsystem. The corresponding maximum storage capacity of the molten
salt tank permits continuous electricity generation by a 50 MW steam
turbine for a duration of 9 h.
2.2. Operating principle
The heated working fluid at the outlet of the hot molten salt tank
further elevates the temperature of the steam entering both the high-
The system configuration of the entire power plant is illustrated in

4
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 3. The photograph of key equipment in the Yumen Xinneng 50 MW beam-down CSP plant.

exchange terminal difference. Additionally, a portion of the third stage

Table 2
extraction is reheated by the hot molten salt and reintroduced into the
Designed parameters of key equipment in the Yumen Xinneng 50 MW
main steam of the LP turbine. Similarly, the first stage extraction of the
beam-down CSP plant.
LP turbine, along with the condensate from #3 HP heater and #3 LP
Parameters Value
heater, are mixed in the deaerator to complete the deaeration process
The entire plant during condensation. The second to fourth stage extractions of the LP
Nominal capacity (MW) 50 turbine engage in heat exchange with the feedwater through #1–3 LP
Power generation (million kW•h) 215
heaters. Finally, the exhaust from the LP turbine is discharged into the
Land area (square kilometers) 3.69
Latitude 40◦ 33′ N condenser to maintain the vacuum within the steam turbine.
Longitude 97◦ 27′ E
Heliostat 3. Methods
Length (m) 4.4
Width (m) 4
Height (m) 2.2 The mathematic models for the key components in the beam-down
Operating temperature (oC) (-40, 65) CSP plant are listed in Section 3.1. The application of the Monte Carlo
Operating relative humidity (%) 5–90 method for simulating the ray tracing process is introduced in Section
Tracking accuracy (mrad) 2
3.2. Furthermore, Section 3.3 lists the performance indices of the entire
Survival wind speed (m/s) 36
Number of heliostats 2603
power plant. Section 3.4 outlines the calculation procedures for the
Radius of heliostat field (m) 225 overall mathematical model of the entire power plant. Lastly, Section
Hyperboloid mirror 3.5 validates the accuracy of the mathematical model by comparing the
Tower height (m) 74 optical efficiency and flux density with existing references.
Hyperboloid mirror height (m) 60–74
Radius (m) 33
Operating temperature (oC) (-40, 150) 3.1. Beam-down plant model
Survival wind speed (m/s) 36
Receiver
Thermal capacity (MWt) 17.1 The mathematic models for the heliostat field, hyperboloid mirror,
Height (m) 11 receiver, molten salt tank, and power cycle are elaborated in Sections
Inlet temperature (oC) 290 3.1.1 to 3.1.5, respectively.
Outlet temperature (oC) 570
Steam turbine
3.1.1. Heliostat field
Annual design utilization hours (h) 4300
Cycle efficiency (%) 43.7 The optical efficiency of the heliostat field for beam-down CSP sys­
Cycle pressure (MPa) 14 tem is defined as Eq. (1):
Molten salt tank
Heat storage time (h) 9 ηhf = ηcos,h × ηsb,h × ηsb,hm × ηat,h × ηat,hm × ηin,hm × ηin,cpc × ρr,h × ρc,h × ρr,hm
Salt type Solar salt
× ρc,hm
(1)
pressure (HP) turbine and low-pressure (LP) turbine through the su­
perheater and reheater, respectively. The hot molten salt facilitates the where ηhf is the optical efficiency of the heliostat field; ηcos,h, ηsb,h, and
heating of water to steam through the evaporator, subsequently raising ηat,h are the cosine efficiency, shading and blocking efficiency, and
the temperature of water entering the evaporator through the preheater, attenuation efficiency of the heliostat, respectively; ηsb,hm, ηat,hm, and ηin,
before ultimately returning to the cold molten salt tank and receiver. In hm are the shading and blocking efficiency, attenuation efficiency, and
the case of the steam turbine, a three-stage extraction system is intercept efficiency of the hyperboloid mirror, respectively; ηin,cpc is the
employed in the HP turbine to heat the feedwater through #1–3 HP intercept efficiency of the CPC; ρr,h and ρc,h are the reflectivity and
heaters, thereby enhancing thermal efficiency by minimizing the heat cleanliness of the heliostat; ρr,hm and ρc,hm are the reflectivity and
cleanliness of the hyperboloid mirror.

5
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 4. Schematic diagram of the Yumen Xinneng 50 MW beam-down CSP plant.

The cosine efficiency of heliostat is calculated by Eq. (2) [17]: reaching the receiver; nl,hm is the number of rays leaving the hyperboloid
mirror.
ηcos,h = cosθin = ̂
ds⋅̂
dn (2)
3.1.2. Hyperboloid mirror
where θin is the angle between the above two vectors, rad; ̂ d s is the The three-dimensional coordinate equation of the hyperboloid
direction vector from the sun to the center of heliostat; d
̂ n is the normal mirror is represented as [7]:
vector of the heliostat. √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( )2
The shading and blocking efficiency of heliostat and hyperboloid zf1 − zf2 x2 + y2hm zf1 − zf2
zhm = + hm2 + (8)
mirror are listed in Eq. (3) and Eq. (4) [18]: 4e 2 e − 1 2
ss,h + sb,h
ηsb,h = 1 − (3) where xhm, yhm, and zhm are the three-dimensional coordinate parame­
sh
ters of the hyperboloid mirror, m; zf1 and zf2 are the heights of the upper
ss,hm focal point and lower focal point, m; e is the eccentricity.
ηsb,hm = 1 − (4)
sh The vertex ratio is defined in Eq. (9) to determine the three-
dimensional shape of a hyperboloid mirror.
where ss,h and sb,h are the shaded area and blocked area by the adjacent
heliostats, respectively, m2; sh is mirror area of a heliostat, m2; ss,hm is
1 chm zmin − zf2
f = + = (9)
the shaded area of a heliostat by the hyperboloid mirror, m2.
2 zf1 − zf1 zf1 − zf2
The attenuation efficiency of heliostat is given in Eq. (5) [19]:
where f is the vertex ratio; chm is the semi-minor axis, m; zmin is the
{
0.99321 − 0.0001176 × d + 1.97 × 10− 8 × d2 (d⩽1, 000 m) lowest height of the hyperboloid mirror, m.
ηat,h = (5) The maximum radius of the hyperboloid mirror is determined by the
e− 0.0001106×d (d > 1, 000 m)
heliostat field radius, which is calculated in Eq. (10) [7]:
where d is the distance from the heliostat to the hyperboloid mirror, m.
zmax − Hh − Lh sinφ zf1 + Lh cosφcotgφ − Hh
Similarly, the value of ηat,hm can be calculated in Eq. (5) based on the = (10)
Rhf − Rhm + Lh cosφ Rhf + Lh cosφ
distance from the hyperboloid mirror to the receiver.
The intercept efficiency of the hyperboloid mirror and CPC are where zmax is the highest height of the hyperboloid mirror, m; Lh and Hh
described in Eq. (6) and Eq. (7) [8]: are the length and height of the heliostat, m; Rhf and Rhm are the
nr,hm maximum radius of the heliostat field and hyperboloid mirror, m; φ is
ηin,hm = (6)
nl,h the zenith angle, rad, which is calculated as follow:
( )
nr,re zf1 − Hh
ηin,cpc = (7) φ = arctg (11)
nl,hm Rhf

where nr,hm is the number of rays reaching the hyperboloid mirror; nl,h is 3.1.3. Compound parabolic concentrator and receiver
the number of rays leaving the heliostat field; nr,re is the number of rays The polygons number of CPC for the Yumen Xinneng beam-down

6
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

CSP plant is 24, which is close to the performance of the ideal CPC [20]. energy of air, J/kg; mi,air is the mass flow rate of air at the inlet, kg/s; hi,
The three-dimensional coordinate equation of CPC is described as [7]: air is the specific enthalpy of air at the inlet, J/kg; Qair is the heat flow of
air, W; ρair is the density of air, kg/m3; Vms is the volume of molten salt,
2 × lfl,cpc sinαsin(β − γ)
xcpc = − ro,cpc sinα (12) m3.
1− β
3.1.5. Power cycle
2 × lfl,cpc cosαsin(β − γ)
ycpc = − ro,cpc cosα (13) The energy balance equations and mass balance equations of the
1− β
superheater, reheater, evaporator, and preheater are shown as follows
2 × lfl,cpc cos(β − γ) [24,25]:
zcpc = (14) ∫ Ti,sh,ms
1− β ( )
Qsh,pc = msh,ms Cp,ms dT = msh,sm × ho,sh,sm − hi,sh,sm (23)
where xcpc, ycpc, and zcpc are the three-dimensional coordinate param­ To,sh,ms

eters of the CPC, m; lfl,cpc is the focal length of the CPC, m, which can be ∫ Ti,rh,ms ( )
calculated in Eq. (15); the range of α is from 0 to 2π; γ is acceptance angle Qrh,pc = mrh,ms Cp,ms dT = mrh,sm × ho,rh,sm − hi,rh,sm (24)
of the CPC, rad; ro,cpc is the radius at the outlet of the CPC, m, which can To,rh,ms

be calculated in Eq. (16). ∫ Ti,ev,ms ( )

lfl,cpc = ro,cpc sin(1 + γ) (15) Qev,pc = mev,ms Cp,ms dT = mev,w × ho,ev,sm − hi,ev,w (25)
To,ev,ms

ro,cpc = ri,cpc sinγ (16) ∫ Ti,ph,ms ( )

Qph,pc = mev,ms Cp,ms dT = mph,w × ho,ev,w − hi,ev,w (26)
where ri,cpc is the radius at the inlet of CPC, m. To,ph,ms

The thermal efficiency of the receiver is defined in Eq. (17) and Eq.
mev,ms = mph,ms = msh,ms + mrh,ms (27)
(18) [21]:
( )
mms Cp,ms To,ms − Ti,ms Qi,re − Qloss,re where Qsh,pc, Qrh,pc, Qev,pc, and Qph,pc are the heat power of superheater,
ηre = = (17) reheater, evaporator, and preheater, respectively, W; msh,pc, mrh,pc, mev,
Qi,re Qi,re
pc, and mph,pc are the mass flow rate of molten salt through the super­
where ηre is the thermal efficiency of the receiver; mms is the mass flow heater, reheater, evaporator, and preheater, respectively, kg/s; Ti,ms and
rate of molten salt, kg/s; Cp,ms is the specific heat of molten salt, J/ To,ms are the molten salt temperature at the inlet and outlet of compo­
(kg•K); To,ms and Ti,ms are the molten salt temperature at the outlet and nent, oC; msm and mw are the mass flow rate of steam and water, kg/s; hi,
inlet of receiver, oC; Qi,re is the input energy of receiver, W; Qloss,re is the sm and ho,sm are the enthalpy of steam at the inlet and outlet of
heat loss of receiver, W, which is calculated in Eq. (18). component, J/kg; hi,w and ho,w are the enthalpy of water at the inlet and
( ) ( ) outlet of component, J/kg; subscript sh, rh, ev, and ph are the super­
heater, reheater, evaporator, and preheater, respectively.
4
Qloss,re = Qcon,re + Qrad,re = hc × sre × Ti,re − T0 + σ × ε × sre × Ti,re − T04
(18) The output work and thermal efficiency of the HP turbine and LP
turbine are listed in Eq. (28) to Eq. (31) [25]:
where Qcon,re and Qrad,re are the convective heat loss and radiative heat
Whpt,pc = mi,hpt,sm hi,hpt,sm − m1e,hpt,sm h1e,hpt,sm − m2e,hpt,sm h2e,hpt,sm − m3e,hpt,sm h3e,hpt,sm
loss of receiver, W; hc is the heat transfer coefficient, W/(m2•K); sre is the
(28)
area of receiver, m2; Ti,re is the temperature at the inlet of receiver, oC; T0
is the ambient temperature, oC; σ is the Stefan-Boltzmann constant, W/ hi,hpt,sm − h1e,hpt,sm
(m2•K4); ε is the emissivity. ηhpt,pc = (29)
hi,hpt,sm − h1e,hpt,sm,s

3.1.4. Molten salt tank Wlpt,pc = mi,lpt,sm hi,lpt,sm − m1e,lpt,sm h1e,lpt,sm − m2e,lpt,sm h2e,lpt,sm − m3e,lpt,sm h3e,lpt,sm
The energy balance equation and mass balance equation of molten − m4e,lpt,sm h4e,lpt,sm − mex,lpt,sm hex,lpt,sm
salt are described in Eq. (19) and Eq. (20), respectively [22]: (30)
( )
d mms Cp,ms Tms
= mi,ms h i,ms − mo,ms h o,ms − Qms + Qreh (19) hi,lpt,sm − h1e,lpt,sm
dt ηlpt,pc = (31)
hi,lpt,sm − h1e,lpt,sm,s
dmms
= mi,ms − mo,ms (20) where Whpt,pc and Wlpt,pc are the output work of the HP turbine and LP
dt
turbine, W; ηhpt,pc and ηlpt,pc are the thermal efficiency of the HP turbine
where Tms and is the molten salt temperature, oC; hi,ms and ho,ms are the and LP turbine; mi,hpt,sm and mi,lpt,sm are the mass flow rate of steam at
specific enthalpy at the inlet and outlet of the molten salt tank, J/kg; Qms the inlet of the HP turbine and LP turbine, kg/s; hi,hpt,sm and hi,lpt,sm are
is the heat flow of molten salt, W; Qreh is the heat flow of molten salt the enthalpy of steam at the inlet of the HP turbine and LP turbine, J/kg;
transferred to ambient, W; mi,ms and mo,ms are the mass flow rate at the m1e,sm, m2e,sm, m3e,sm, m4e,sm, and mex,sm are the mass flow rate of first
inlet and outlet of the molten salt tank, kg/s. stage extraction, second stage extraction, third stage extraction, fouth
The energy balance equation and mass balance equation of air are stage extraction, and exhaust J/kg; h1e,sm, h2e,sm, h3e,sm, h4e,sm, and hex,sm
described in Eq. (21) and Eq. (22), respectively: are the enthalpy of first stage extraction, second stage extraction, third
stage extraction, fouth stage extraction, and exhaust, J/kg; h1e,sm,s is the
d(mair uair )
= mi,air h i,air − Qair (21) isentropic enthalpy, J/kg; subscript hpt and lpt are the HP turbine and
dt
LP turbine.
( )
dmair dVms
= − ρair (22) 3.2. Ray tracing method
dt dt

where mair is the mass flow rate of air, kg/s; uair is the specific internal Based on the statistical methods, the Monte Carlo ray tracing method
is applied in this research to analyze the reflection and absorption

7
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

processes of sunlight on the heliostat field, hyperboloid mirror, CPC, and nh

∑
receiver [27]. The equations for incident sunlight are described in Eq. Qhf = DNI × sh,i × ηhf,i (44)
(32) to Eq. (34) [6]: i=1

xir,h = xi,h + k⋅ ̂
d s,x,h (32) where Qhf is the heliostat field capacity, W; DNI is the direct normal
irradiation, W/m2; sh,i is the mirror area of the i-th heliostat, m2; ηhf,i is
yir,h = yi,h + k⋅ ̂
d s,y,h (33) the optical efficiency of the i-th heliostat, which is calculated in Eq. (1);
nh is the number of heliostat.
̂ s,z,h
zir,h = zi,h + k⋅ d (34)
(2) Hyperboloid mirror and receiver
where xir,h, yir,h, and zir,h are the coordinates of incident sunlight; xi,h, yi,
h, and zi,h are the coordinates of the i-th sunlight on the surface of he­ The flux density of hyperboloid mirror and receiver is defined as the
liostat; ̂
d s,x,h , ̂
d s,y,h , and ̂
d s,z,h are the components of direction vector from energy of sunlight within each grid [29], shown in Eq. (45):
the sun to the center of heliostat. nsl,g × Esl
The normal of the intersection point between the sunlight reflected I= (45)
sg
by the heliostat and hyperboloid mirror are as follows:
where I is the flux density, W/m2; nsl,g is the number of sunlight within
̂ n,x,hm = ⃦∂fhm /∂⃦x
d (35) each grid; sg is the mirror area of the grid, m2; Esl is the energy of sun­
⃦̂d n,hm ⃦
light, W, which is calculated in Eq. (46).

̂ n,y,hm = ⃦∂fhm /∂⃦y

∑
nh
d (36) Esl = (DNI × ηcos,h,i × sh,i )/nsl,t (46)
⃦̂d n,hm ⃦ i=1

̂ n,z,hm = ⃦∂fhm /∂⃦z where ηcos,h,i is the cosine efficiency of the i-th heliostat; nsl,g is the total
d (37)
⃦̂d n,hm ⃦ number of sunlight.
For the molten salt tank, the heat storage of the molten salt tank is
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
)̅
( ) ( ) ( listed in Eq. (19) to Eq. (22), and the output work of the steam turbine is
⃦ ⃦ ∂fhm 2 ∂fhm 2 ∂fhm 2
⃦̂d n,hm ⃦ = + + (38) defined in Eq. (28) and Eq. (31).
∂x ∂y ∂z
3.4. Theoretical solution procedures
where d ̂ n,x,hm , ̂
d n,y,hm , and d
̂ n,z,hm are the components of normal vector;
fhm is the equation function of hyperboloid mirror in Eq. (9); ̂ d n,hm is the The model calculation procedures for the mathematical model of the
normal of the intersection point. entire power plant are illustrated in Fig. 5. The model calculation in­
The coordinates of the ray reflected by the hyperboloid mirror are cludes the following four processes, which are Monte Carlo submodel,
shown in Eq. (39) to Eq. (41): optical efficiency analysis submodel, molten salt tank submodel, and
( ) power cycle submodel. Firstly, the coordinates of intersection points are
̂ s,x,h − 2⋅ d
xrr,hm = d ̂ n,x,hm ̂ d n,y,hm ̂
d s,x,h + ̂ d s,y,h + ̂ ̂ s,z,h ⋅d
d n,z,hm d ̂ n,x,hm (39)
calculated based on the Monte Carlo ray tracing method to conduct the
( ) flux density simulation of the hyperboloid mirror and receiver. Sec­
̂ s,x,h − 2⋅ d
yrr,hm = d ̂ n,x,hm ̂ d n,y,hm ̂
d s,x,h + ̂ d s,y,h + ̂ ̂ s,z,h ⋅ ̂
d n,z,hm d d n,y,hm (40)
ondly, the shaded area and blocked area of each heliostat and the
( ) number of intercepted rays are calculated to obtain the values of ηsb,h,
zrr,hm = ̂
d s,x,h − 2⋅ d̂ n,x,hm ̂
d s,x,h + ̂ ̂ s,y,h + d
d n,y,hm d ̂ n,z,hm ̂ ̂ n,z,hm
d s,z,h ⋅ d (41) ηsb,hm, ηin,hm, and ηin,cpc. Thirdly, the energy and mass balance equations
of the two molten salt tanks are solved to obtain the heat flow during the
where xrr,hm, yrr,hm, and zrr,hm are the coordinates of reflected sunlight. charging and discharging processes. Finally, the energy analysis of the
The solar elevation angle and solar azimuth angle are applied to superheater, reheater, evaporator, preheater, and steam turbine is car­
describe the position of the sun, which are listed in Eq. (42) and Eq. (43) ried out to evaluate the performance of power generation in this
[28]. research.
asea = arcsin(sinϕsinδ + cosϕcosδcosω) (42)
⃒ ( )⃒ 3.5. Model validation
⃒ sinasea sinϕ − sinδ ⃒⃒
asaa = sgn(ω)⃒⃒arccos (43)
cosasea cosϕ ⃒ To validate the accuracy of the mathematical models of heliostats
based on the Monte Carlo ray tracing method, a comparison between the
where asea and asaa are the solar elevation angle and solar azimuth angle, optical efficiency of heliostats obtained from reference [8] and simula­
◦
; ϕ is the latitude; δ is the declination angle; ω is the solar hour angle. tion results is presented in Fig. 6. The design parameters of the virtual
beam-down CSP plant are listed in Table 3, situated at a latitude of
3.3. Performance indices 23.8◦ S and a longitude of 133.9◦ E. The heliostat field comprises 2,181
heliostats, each with an area of 50 m2. The design heights of the hy­
The mathematical model of performance indices is mainly used to perboloid mirror and receiver are 50 m and 10 m, respectively. The
evaluate the operating performance of key components in the entire heliostat field efficiency and flux density distribution are analyzed based
power plant, shown in Eq. (44) to Eq. (46): on the Monte Carlo ray tracing method at 12:00 on March 21st. The
variation of cosine efficiency and blocking efficiency of heliostats under
(1) Heliostat field: various coordinates of xh and yh is illustrated in Fig. 6(b). The co­
ordinates of xh and yh range from 4.5 to 438. The overall trend of optical
In addition to using the optical efficiency of the heliostat field in Eq. efficiency of heliostats drops with increasing distance from the origin.
(1), the heliostat field capacity is defined in Eq. (44): Correspondingly, the optical efficiency of heliostats located in the first
and second quadrants (zone 1) is higher than that of heliostats located in
the third and fourth quadrants (zone 2), with the reason that the cosine

8
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 5. The calculation procedures for the mathematical model of the entire power plant.

9
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 6. Comparison of optical efficiency of heliostats between reference [8] and simulation results.

hyperboloid mirror is determined to be 33.8 kW/m2. Furthermore, the

Table 3 simulation results of receiver flux density are analyzed in this section. It
Designed parameters of the virtual beam-down CSP plant in reference [8].
is evident that the highest flux density of the receiver is observed at the
Parameters Value origin, owing to the concentrating effect of the hyperboloid mirror and
The entire plant CPC. The average flux density values of the receiver within the radius of
Nominal capacity (MW) 10 3.4 m, 4.2 m, and 5 m are determined to be 460.5 kW/m2, 414.4 kW/m2,
Solar multiple 1 and 373.9 kW/m2, respectively. Correspondingly, the CPC intercept
Direct normal irradiation (W/m2) 950
Latitude 23.8◦ S
efficiency values within the same radii are 0.83, 0.91, and 0.95. The
Longitude 133.9◦ E reason is that the submirror of the hyperboloid mirror is not an ideal
Heliostat surface, resulting in certain optical errors. The similar color distribution
Number of heliostats 2,181 of the hyperboloid mirror and receiver between reference [8] and
Area (m2) 50
simulation results indicates the effectiveness of the Monte Carlo sub­
Height (m) 2.2
Radius of heliostat field (m) 438 model. This comparison further validates the accuracy of the mathe­
Hyperboloid mirror matical models employed in the study.
Tower height (m) 50
Vertex ratio 0.85 4. Case study
Receiver
Height (m) 10
Operating temperature (K) 800–1000 The Yumen Xinneng 50 MW beam-down CSP plant recently under­
Thermal storage medium Solid particles went a technical transformation at the end of May, which involved
Efficiency (%) 75 replacing the hyperboloid mirror with a higher reflectivity. With respect
Power cycle
to the operational results of module #4, the variation of direct normal
Types Brayton cycle
Efficiency (%) 50 irradiation (DNI), wind speed, number of heliostats (nh), inlet temper­
ature of molten salt (Ti,ms), and outlet temperature of molten salt (To,ms)
is depicted in Fig. 8 and Fig. 9. Taking the June 12th as an example, the
efficiency of zone 1 is larger than that of zone 2. The corresponding DNI increased from 667 W/m2 at 10:45 to the maximum value of 753
average optical efficiency values for zone 1 and zone 2 are 0.72 and W/m2 at 13:31, followed by a gradual decline to 577 W/m2 at 18:00.
0.56, respectively. Furthermore, the optical efficiency of point 1 is The fluctuation of wind speed had a significant impact on the intercept
different from that of point 2 on the same circle. The reason is that the efficiency of the hyperboloid mirror and CPC. The wind speed values
shaded area and blocked area of point 1 are significantly larger than that varied in the range of 3.5 m/s to 11.6 m/s, which were below the sur­
of point 2, though the cosine efficiency values of point 1 and point 2 are vival wind speed of 36 m/s.
similar. The corresponding shaded area and blocked area of point 1 and For stage 1, the outlet temperature of molten salt increased from
point 2 are 12 m2 and 0 m2, respectively. The similar color distribution 384.9◦ C at 10:45 to 439.9◦ C at 11:36, owing to a rise in the number of
of heliostats in Fig. 6 indicates the effectiveness of the Monte Carlo heliostats from 1,990 to 2,035. The corresponding inlet temperature of
submodel and optical efficiency analysis submodel. molten salt ranged between 301.6◦ C and 301.9◦ C. For stage 2 between
As presented in Fig. 7, the flux density of the hyperboloid mirror and 11:36 and 11:50, the number of heliostats decreased from 2,035 to
receiver is compared between reference [8] and simulation results. In 1,789, resulting in a corresponding decrease in the outlet temperature of
Fig. 7(b), it is observed that the flux density of the hyperboloid mirror is molten salt from 439.9◦ C to 411.8◦ C. The inlet temperature of molten
larger in zone 1 compared to zone 4. The difference can be attributed to salt remained in the range of 301.5◦ C to 301.6◦ C during stage 2. With
the higher optical efficiency of the heliostat in zone 1 compared to zone respect to stage 3 between 11:50 and 12:51, the outlet temperature of
4. In zone 2, a portion of the hyperboloid mirror area has a flux density molten salt increased from 411.8◦ C at 2,173 heliostats to 528◦ C at 2,311
of 0 kW/m2 for the reason that the corresponding heliostats are obscured heliostats. The corresponding inlet temperature of molten salt ranged
by the shadow of the hyperboloid mirror. The average flux density of the

10
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 7. Comparison of hyperboloid mirror and receiver flux density between reference [8] and simulation results.

between 301.2◦ C and 301.5◦ C. For the purpose of preventing over­ photos of the receiver are presented in Fig. 9(d) and Fig. 9(e). The CPC
heating, the number of heliostats decreased from 2,311 at 12:51 to 1,388 has the effect of the secondary concentration of rays reflected by the
at 13:18, then gradually increased to the maximum value of 2,316 at hyperboloid mirror. Furthermore, a comparison between the light spot
13:47, before dropping to 1,808 at 14:05. The corresponding outlet on the receiver and the black background demonstrates that the over­
temperature of molten salt increased from 528◦ C at 12:58 to 558◦ C at flow loss in the Yumen Xinneng 50 MW beam-down CSP plant was
13:07 due to thermal inertia. The average outlet temperature of molten relatively low. Overall, the technical transformation of the entire power
salt was 559◦ C between 13:07 and 13:56. The reduction in the number plant has yielded promising outcomes, and the operational results of
of heliostats caused the outlet temperature of molten salt to drop from module #4 demonstrate the effectiveness of the new hyperboloid mirror
558.4◦ C at 13:56 to 542.7◦ C at 14:05. In stage 5, the DNI decreased from with higher reflectivity.
740 kW/m2 at 14:05 to 682 kW/m2 at 16:37, prompting an increase in To further conduct a comprehensive performance analysis, the
the number of heliostats from 1,808 to 2,126 to enhance the outlet theoretical evaluation of the heliostat field, hyperboloid mirror,
temperature of molten salt. The corresponding outlet temperature of receiver, and power cycle is discussed in Section 5.1 to 5.4.
molten salt dropped from 542.7◦ C at 14:05 to the minimum value of
506.42◦ C at 15:05, before gradually increasing to 540.9◦ C at 16:37. For 5. Results and discussion
stage 6 between 16:37 and 18:00, the outlet temperature of molten salt
decreased from 540.9◦ C at DNI of 683 W/m2 to 515◦ C at DNI of 577 W/ As depicted in Section 5, the variation of the three-dimensional co­
m2. The corresponding number of heliostats varied between 2,095 and ordinates of the hyperboloid mirror with eccentricity and vertex ratio is
2,149. Additionally, the inlet temperature of molten salt increased from analyzed in Section 5.1. The simulation of the cosine efficiency, shading
302.3◦ C at 15:00 to 307.6◦ C at 18:00, as a result of the rewarming effect and blocking efficiency, shading efficiency of the hyperboloid mirror,
caused by the increase in hot molten salt. and attenuation efficiency for the heliostat field is conducted in Section
The photos of the hyperboloid mirror and receiver in operation are 5.2. Furthermore, the evaluation of the flux density distribution of the
described in Fig. 9. Fig. 9(a) provides a visual representation of the hyperboloid mirror and receiver under four typical days is discussed in
hyperboloid mirror prior to light concentration. The color distribution of Section 5.3. Finally, Section 5.4 evaluates the annual performance of the
the hyperboloid mirror under 1,388 heliostats at 13:18 is depicted in Yumen Xinneng 50 MW power plant.
Fig. 9(b). Similarly, Fig. 9(c) showcases the color distribution of the
hyperboloid mirror with 2,301 heliostats at 13:43. The experimental

11
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 8. The variation of operational results on June 12th.

Fig. 9. The photos of the hyperboloid mirror and receiver in operation.

12
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

5.1. Optimized hyperboloid mirror parameters

The shape of the hyperboloid mirror has an impact on the optical and
thermal performance of the beam-down CSP plant, especially the flux
density and the area of the hyperboloid mirror and receiver, which in
turn affects the initial investment. Consequently, it is crucial to analyze
the he variation of xhm, yhm, and zhm with eccentricity and vertex ratio
under a heliostat field radius of 225 m in Fig. 10 and Fig. 11. As
expressed in Eq. (8) and Eq. (9), the vertex ratio values decrease from 0.9
to 0.505 as the eccentricity increases. Consequently, the height of the
hyperboloid mirror decreases with increasing eccentricity, leading to an
increase in both area and radius. The minimum values of zhm corre­
sponding to eccentricities of 1.25, 2.5, 5, 10, 20, 50, and 100 are 73.1 m,
59.3 m, 52.4 m, 48.95 m, 47.225 m, 46.19 m, and 45.845 m, respec­
tively. Additionally, the corresponding maximum values of zhm reduce
from 77.84 m at an eccentricity of 1.25 to 47.29 m at an eccentricity of
100. Conversely, the maximum values of xhm increase from 12.38 m at Fig. 11. The effect of vertex ratio on the Rre, Shm, and Rhm.
an eccentricity of 1.25 to 100.73 m at an eccentricity of 100. Further­
more, the difference in height between maximum and minimum values June 1st, September 1st, and December 1st are expressed in Fig. 12 and
of zhm drops from 4.74 m at an eccentricity of 1.25 to 1.44 m at an ec­ Appendix Figs. A1-A3, which consist of cosine efficiency, shading and
centricity of 100. As depicted in Fig. 10(b), the coordinates of the blocking efficiency, shading efficiency of hyperboloid mirror, and
maximum values of xhm and zhm exhibit a linear relationship based on attenuation efficiency. The 2,603 heliostats in each heliostat field are
the calculation results outlined in Eq. (10). classified into three regions: (a) the north sector consisting of 869 he­
The effect of the vertex ratio on the receiver radius, hyperboloid liostats, (b) the west sector including 867 heliostats, and (c) the east
mirror area, and hyperboloid mirror radius is illustrated in Fig. 11. It can sector including 867 heliostats. The 2,603 heliostats are distributed in
be observed that the hyperboloid mirror area decreases from 5628.9 m2 24 circles across the three sectors, with the corresponding circle spacing
at a vertex ratio of 0.7 to 558.7 m2 at a vertex ratio of 0.9. The reduction increasing from 6.1 m in the 1st cycle to 12.2 m in the 23rd cycle. The
in area is accompanied by a corresponding decrease in the hyperboloid radius of the heliostat field varies in the range of 42.8 m to 225 m. Since
mirror radius from 38.6 m at a vertex ratio of 0.7 to 12.38 m at a vertex the tower blocks the reflected rays from the heliostat field, the angles of
ratio of 0.9. The light spot on the receiver expands with the increase of the boundary lines between the north and east sectors are 19.52◦ and
the hyperboloid mirror area, leading to an increase in the receiver radius 37.6◦ , respectively. Similarly, the angles of boundary lines between the
from 3.16 m at a vertex ratio of 0.7 to 10.15 m at a vertex ratio of 0.9. north and west sectors are 142.4◦ and 160.48◦ , respectively, and that of
Nevertheless, a hyperboloid mirror area of 5628.9 m2 at a vertex ratio of boundary lines between the west and east sectors are 262.42◦ and
0.7 necessitates a tower with higher strength when compared to a vertex 277.58◦ , respectively. These sector divisions enable the analysis of the
ratio of 0.9. Consequently, the height of the hyperboloid mirror under a heliostat field efficiency in different regions.
vertex ratio of 0.7 exhibits lower cosine efficiency relative to that under On March 1st in Fig. 12(a), the average cosine efficiency in the north
a vertex ratio of 0.9. The recommended design zone in this research is sector is higher compared to the west and east sectors. The corre­
between 0.775 and 0.8 for mitigating the contradiction between initial sponding average values of cosine efficiency for the north sector and the
investment and optical efficiency, providing a suitable compromise for entire heliostat field are 96.89% and 80.43%, respectively. The
the design of the hyperboloid mirror. discrepancy can be attributed to the angle calculation in Eq. (2), where
The vertex ratio of the hyperboloid mirror is selected as 0.775 to the angle for the north sector is smaller than that for the west and east
analyze the theoretical performance of the entire beam-down CSP plant sectors when the sun is positioned on the south side of the heliostat field.
in Section 5.2 to 5.4. Consequently, the north sector exhibits a higher cosine efficiency.
Furthermore, the average cosine efficiency in the west and east sectors at
5.2. Heliostat field efficiency 12:00 is similar, with the reason that the azimuth angle of the sun is
close to the south direction. The corresponding average values of cosine
The simulation results of the heliostat field efficiency on March 1st, efficiency for the west and east sectors are 70.61%, and 73.74%,

Fig. 10. The variation of xhm, yhm, and zhm with eccentricity and vertex ratio.

13
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 12. Simulation results of heliostat field efficiency on March 1st.

respectively. For the north sector, the average efficiency increases from 190.52◦ , respectively. The shaded and blocked areas on June 1st and
97.31% in the 1st cycle to a maximum value of 98.13% in the 3rd cycle, September 1st are primarily located in the north and east sectors, while
and then decreases to a minimum value of 95.53% in the 24th cycle. As those on March 1st and December 1st are predominantly situated in the
illustrated in Appendix Figs. A1-A3(a), the average values of cosine ef­ west and east sectors. It can be concluded that the location of the shaded
ficiency on June 1st, September 1st, and December 1st are 85.56%, and blocked areas rotates in a clockwise manner as the solar azimuth
83.72%, and 76.32%, with the reason that solar elevation angle on angle increases.
March 1st, June 1st, September 1st, and December 1st is 41.85◦ , 70.72◦ , The variation of shading efficiency of the hyperboloid mirror under
57.38◦ , and 27.22◦ , respectively. The results highlight the effect of solar four typical days is expressed in Fig. 12(c) and Appendix Figs. A1 to A3
elevation angle on the overall cosine efficiency of the heliostat field. (c), which is a key characteristic distinguishing it from conventional CSP
Regarding the shading and blocking efficiency on March 1st, it is plants. The distance between the shaded area and the origin is influ­
observed that the shaded and blocked areas are primarily distributed in enced by the solar elevation angle, resulting in variations in shading
the west and east sectors for the reason that the sun is close to the south efficiency. On March 1st, June 1st, September 1st, and December 1st, the
orientation at 12:00. The average values of shading and blocking effi­ shadow area positions range from the 2nd to 11th cycles, 1st to 3rd
ciency for the north, west, and east sectors are 99.27%, 98.46%, and cycles, 1st to 6th cycles, and 12th to 18th cycles, respectively. The
98.14%, respectively. Comparing the simulation results with those corresponding average shading efficiency values are determined to be
presented in Fig. 12(b) and Appendix Figs. A1 to A3(b), the shading and 97.25%, 99.35%, 98.14%, and 97.89%, respectively. Regarding the
blocking efficiency reduces as the solar elevation angle increases. Spe­ attenuation efficiency in Fig. 12(d) and Appendix Figs. A1 to A3(d), the
cifically, the average shading and blocking efficiency on June 1st, average values across the four typical days are calculated to be 97.46%
September 1st, and December 1st are 99.96%, 99.76%, and 92.96%, using Eq. (5). It is observed that the average attenuation efficiency re­
respectively. Furthermore, the solar azimuth angle on March 1st, June duces from 98.44% in 1st cycle to 96.66% in 24th cycle.
1st, September 1st, and December 1st is 185.56◦ , 202.25◦ , 193.36◦ , and The simulation results demonstrate the impact of seasonal changes

14
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

on the heliostat field efficiency. The average cosine efficiency is highest insights into the performance of the heliostat field under different sea­
on June 1st and lowest on December 1st, with the shading and blocking sonal conditions, enabling the optimization of the heliostat design and
efficiency exhibiting a similar trend. The results provide valuable operation.

Fig. 13. Simulation results of hyperboloid mirror flux density.

15
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 13. (continued).

5.3. Flux density distribution yre > 0 and yre < 0 is 376.3 kW/m2 and 330.9 kW/m2, respectively. The
corresponding receiver flux density at yre = 3.75 and yre = -3.75 under a
The evaluation of flux density distribution on four typical days is constant xre of 0 is 183.4 kW/m2 and 99.3 kW/m2, respectively. The
conducted in Section 5.3. The variation of hyperboloid mirror flux discrepancy arises from the fact that the hyperboloid mirror flux density
density distribution and receiver flux density distribution with two- of the north sector is greater than that of the west and east sectors in
dimensional and three-dimensional coordinates is described in Fig. 13 Fig. 13(a), resulting in more rays reaching the north zone of the receiver.
and Fig. 14. With respect to the hyperboloid mirror, the flux density As described in Fig. 14(b)-(d), the maximum flux density of the hyper­
distribution is categorized into the north sector, west sector, and east boloid mirror at the origin is observed to be 953.3 kW/m2, 892.6 kW/
sector. The division is a result of the absence of reflected light in the m2, and 701.5 kW/m2 on June 1st, September 1st, and December 1st,
corresponding heliostat field regions. Taking March 1st as an example, respectively. The corresponding average flux density of the receiver on
the maximum flux density in the north, west, and east sectors is 21 kW/ these three days is calculated as 399.1 kW/m2, 381.7 kW/m2, and 314.2
m2, 16 kW/m2, and 18 kW/m2, respectively. The average values of kW/m2, respectively.
hyperboloid mirror flux density for the north, west, and east sectors are
found to be 12.58 kW/m2, 8.74 kW/m2, and 9.3 kW/m2, respectively.
The main reason for higher flux density in the north sector is that the 5.4. Annual performance
cosine efficiency of 96.89% in the north sector is larger than that of
70.61% and 73.74% in the west and east sectors in Fig. 12(a). The The variation of heliostat field efficiency (ηhf), heliostat field ca­
average hyperboloid mirror flux density increases from 10.2 kW/m2 on pacity (Qhf), receiver capacity (Qre), heat storage of molten salt tank
March 1st to 11.2 kW/m2 on June 1st and 10.9 kW/m2 on September 1st (Qmst), and power generation (Wpc) with DNI and wind speed is illus­
due to an increase in the solar elevation angle, while that is 9 kW/m2 on trated in Fig. 15 and Appendix Figs. A4 to A6. As described in Fig. 15, the
December 1st for a lower solar elevation angle. Additionally, the performance of the Yumen Xinneng 50 MW power plant is evaluated
maximum flux density on the four typical days is recorded as 21 kW/m2, from March 1st to March 7th, in which sunny weather is observed from
23 kW/m2, 21 kW/m2, and 22 kW/m2, respectively. The position of the March 1st to March 6th and cloudy weather is experienced on March
fragmentary flux density on the hyperboloid mirror generally corre­ 7th. On March 1st, the DNI ranges from 0 W/m2 to 574 W/m2, and the
sponds to the two-dimensional coordinates of the shaded area in Fig. 12 wind speed fluctuates between 0.3 m/s and 4.7 m/s. The heliostat field
(c). The height of the fragmentary flux density varies within the range of efficiency is maintained at 0 between 1:00–7:00 and 19:00–24:00 due to
65.33 m-67.43 m, 64.82 m-65.24 m, 64.82 m-66.09 m, and 67.72 m- the solar elevation angle being less than 0, then increases from 47.87%
69.33 m on March 1st, June 1st, September 1st, and December 1st, at 8:00 to a maximum value of 52.46% at 13:00, before dropping to
respectively. Furthermore, the corresponding area of the fragmentary 46.71% at 18:00. The variation in heliostat field capacity and receiver
flux density in the north sector on the four typical days is 59.71 m2, capacity is primarily dependent on the fluctuation of DNI due to the
20.91 m2, 49.75 m2, and 25.5 m2, respectively. small change in heliostat field efficiency. Under the operation of 39,045
The flux density distribution at various coordinates of xre and yre on heliostats (15 heliostat fields), the heliostat field capacity and receiver
March 1st, June 1st, September 1st, and December 1st is depicted in capacity increase from 41.27/29.39 MWt at 8:00 to maximum values of
Fig. 14. On March 1st, the receiver flux density reduces from 803.9 kW/ 186.79/174.91 MWt at 14:00, before decreasing to 101.79/89.91 MWt
m2 at xre of 0 to 160.2 kW/m2 at xre of 3.75 under a constant yre of 0. The at 17:00. The operation strategy of this power plant involves prioritizing
reduction follows an approximate Gaussian distribution. The average the storage of hot molten salt in the molten salt tank, with excess heat
receiver flux density within radii of 1 m, 2 m, 3 m, and 3.75 m is being utilized for the power cycle. The heat storage of the molten salt
calculated as 462.9 kW/m2, 420.7 kW/m2, 373.4 kW/m2, and 353.6 tank increases from 27.44 MWt to the designed capacity of 1,125MWt
kW/m2, respectively. Additionally, the average receiver flux density for between 16:00–17:00, and the corresponding power generation remains
at 0 between 3:00–15:00. The values of power generation are 33.7 MWe

16
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 14. Simulation results of receiver flux density.

at 16:00 and 36.86 MWe at 17:00 by utilizing the excess heat from the For the cloudy weather on March 7th, the DNI fluctuates between 0 and
molten salt. Subsequently, power generation remains at 50 MWe be­ 701 W/m2, while the wind speed ranges from 0.5 m/s to 6.8 m/s. The
tween 18:00 on March 1st and 02:00 on March 2nd based on the heat maximum values of heliostat field efficiency, heliostat field capacity,
storage capacity of the molten salt tank. receiver capacity, and heat storage of the molten salt tank are 52.07%,
With respect to the period from March 2nd to March 6th, the trends 221.83 MWt, 209.95 MWt, and 998.76 MWt, respectively. The corre­
in heliostat field efficiency, heliostat field capacity, and power genera­ sponding power generation and duration time are 400 MWe and 8 h,
tion are similar to those observed on March 1st. The values of DNI and respectively.
wind speed during this period range from 0 W/m2 and 0.3 m/s to 911 Similarly, the performance of the Yumen Xinneng 50 MW power
W/m2 and 7.6 m/s, respectively. The maximum values of heliostat field plant in June, September, and December is presented in Appendix
efficiency from March 2nd to March 6th are 52.51%, 51.42%, 52.62%, Figs. A4 to A6. A comparison between the simulation results in June and
52.67%, and 52.02%, respectively. The lower heliostat field efficiency March reveals that the increase in power generation is primarily
on March 3rd is attributed to the high wind speed of 7.6 m/s, which attributed to the extended sunshine duration and the improved heliostat
leads to a reduced intercept efficiency of the hyperboloid mirror and field efficiency. The average sunshine duration in June is 4 h longer than
CPC. The receiver capacity reaches the designed value of 225 MWt from that in March, resulting in a 2% increase in heliostat field efficiency
March 3rd to March 6th, while the maximum value of receiver capacity compared to March. The power generation during the periods of June
on March 2nd is 220.31 MWt due to relatively low DNI. The duration of 1st to 7th, September 1st to 7th, and December 1st to 7th amounts to
the designed receiver capacity and molten salt tank is primarily 4674.22 MWe, 4577.95 MWe, and 4148.72 MWe, respectively. The
dependent on the DNI, in which the duration time on March 4th is 7 h corresponding duration time for a 7-day period in March, June,
and 3 h longer than that on March 1st, respectively. Consequently, the September, and December is 89 h, 100 h, 94 h, and 88 h, respectively.
corresponding power generation and duration time increase from
520.56 MWe and 11 h on March 1st to 700 MWe and 14 h on March 4th.

17
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. 15. Performance evaluation of the Yumen Xinneng 50 MW power plant in March.

6. Conclusions layout and the stability of the molten salt outlet temperature to propose
new operational strategies for enhancing the comprehensive perfor­
This research presents a theoretical and run-test evaluation of the mance of the Yumen Xinneng 50 MW beam-down CSP plant.
Yumen Xinneng 50 MW beam-down CSP plant, aiming to describe the
operating performance of the heliostat, hyperboloid mirror, receiver, CRediT authorship contribution statement
and power cycle. The key conclusions of this paper are described as
follows: Xinzhuang Gu: Writing – original draft, Validation, Software,
Methodology, Investigation. Hao Chen: Validation, Methodology.
(1) The recommended design parameters for the vertex ratio of the Dequan Xu: Software, Methodology. Shixiong Song: Visualization,
hyperboloid mirror in this research are between 0.775 and 0.8 for Methodology. Wentao Xie: Investigation, Funding acquisition. Yuda
mitigating the contradiction between initial investment and op­ Chen: Methodology, Investigation. Raúl Navío Gilaberte: Software,
tical efficiency. Investigation. Teng Jia: Investigation, Funding acquisition. Yanjun
(2) The maximum values of cosine efficiency, shading and blocking Dai: Writing – review & editing, Supervision. Bo Yu: Methodology,
efficiency, shading efficiency of the hyperboloid mirror, and Investigation. Shuochen Zhou: Methodology, Investigation.
attenuation efficiency are 85.56%, 99.96%, 99.35%, and 97.46%
on June 1st for the solar elevation angle reaches up to 70.72◦ .
(3) The average hyperboloid mirror flux density and receiver flux Declaration of competing interest
density on March 1st, June 1st, September 1st, and December 1st
are 10.2/11.2/10.9/9 kW/m2 and 353.6/399.1/381.7/314.2 The authors declare that they have no known competing financial
kW/m2, respectively. interests or personal relationships that could have appeared to influence
(4) The power generation during the periods of March 1st to 7th, the work reported in this paper.
June 1st to 7th, September 1st to 7th, and December 1st to 7th
amounts to 4377.56 MWe, 4674.22 MWe, 4577.95 MWe, and Data availability
4148.72 MWe, respectively.
(5) The run-test results reveal that the maximum outlet temperature No data was used for the research described in the article.
of the molten salt is maintained at 559◦ C for 50 min at an average
DNI of 739.7 W/m2. Acknowledgement

This research contributes to the theoretical and run-test evaluation of The authors thank the National Key R&D Program of China (No.
the Yumen Xinneng 50 MW beam-down CSP plant and proves the 2022YFE0196500), the Key Scientific and Technological Projects 2023
feasibility of commercial operation. The research method can be adop­ of China Power Engineering Consulting Group (No. DG3-A03-2023), the
ted to evaluate the performance of the beam-down CSP plant under Science and Technology Commission of Shanghai Municipality (No.
various seasonal weather conditions in other countries. The future work 23DZ1201000), and the National Key Research and Development Pro­
based on this study would focus on the optimization of the heliostat field gram of China (No. 2023YFB4204301 and 2023YFB4204303).

18
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Appendix

Fig. A1. Simulation results of heliostat field efficiency on June 1st.

19
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. A2. Simulation results of heliostat field efficiency on September 1st.

20
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. A3. Simulation results of heliostat field efficiency on December 1st.

21
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. A4. Performance evaluation of the Yumen Xinneng 50 MW power plant in June.

22
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. A5. Performance evaluation of the Yumen Xinneng 50 MW power plant in September.

23
X. Gu et al. Energy Conversion and Management 308 (2024) 118389

Fig. A6. Performance evaluation of the Yumen Xinneng 50 MW power plant in December.

References [16] M.T. Resource. Standard map service. http://bzdt.ch.mnr.gov.cn/. 2023 [accessed
3 January 2024].
[17] Noone CJ, Torrilhon M, Mitsos A. Heliostat field optimization: a new
[1] Yang XJ, Hu H, Tan T, Li J. China’s renewable energy goals by 2050. Environm Dev
computationally efficient model and biomimetic layout. Sol Energy 2012;86:
2016;20:83–90.
792–803.
[2] Romero M, Steinfeld A. Concentrating solar thermal power and thermochemical
[18] Eddhibi F, Ben Amara M, Balghouthi M, Guizani A. Design and analysis of a
fuels. Energ Environ Sci 2012;5:9234–45.
heliostat field layout with reduced shading effect in southern Tunisia. Int J
[3] Müller-Steinhagen H, Trieb F. Concentrating solar power, a review of the
Hydrogen Energy 2017;42:28973–96.
technology. Ingenia Inform QR Acad Eng 2004;18:43–50.
[19] Schmitz M, Schwarzbözl P, Buck R, Pitz-Paal R. Assessment of the potential
[4] Z. Wang. The Blue Book of China’s Concentrating Solar Power Industry 2022.
improvement due to multiple apertures in central receiver systems with secondary
www.cnste.org. 2022 [accessed 3 January 2024].
concentrators. Sol Energy 2006;80:111–20.
[5] Rabl A. Tower reflector for solar power plant. Sol Energy 1976;18:269–71.
[20] Cooper T, Dähler F, Ambrosetti G, Pedretti A, Steinfeld A. Performance of
[6] Segal A, Epstein M. The optics of the solar tower reflector. Sol Energy 2001;69:
compound parabolic concentrators with polygonal apertures. Sol Energy 2013;95:
229–41.
308–18.
[7] Leonardi E. Detailed analysis of the solar power collected in a beam-down central
[21] Ali M, Rady M, Attia MAA, Ewais EMM. Consistent coupled optical and thermal
receiver system. Sol Energy 2012;86:734–45.
analysis of volumetric solar receivers with honeycomb absorbers. Renew Energy
[8] Saldivia D, Bilbao J, Taylor RA. Optical analysis and optimization of a beam-down
2020;145:1849–61.
receiver for advanced cycle concentrating solar thermal plants. Appl Therm Eng
[22] Tagle-Salazar PD, Prieto C, López-Román A, Cabeza LF. A transient heat losses
2021;197:117405.
model for two-tank storage systems with molten salts. Renew Energy 2023;219:
[9] Wei X, Lu Z, Yu W, Xu W. Ray tracing and simulation for the beam-down solar
119371.
concentrator. Renew Energy 2013;50:161–7.
[24] Al Kindi AA, Sapin P, Pantaleo AM, Wang K, Markides CN. Thermo-economic
[10] Grange B, Kumar V, Gil A, Armstrong PR, Codd DS, Slocum A, et al. Preliminary
analysis of steam accumulation and solid thermal energy storage in direct steam
optical, thermal and structural design of a 100 kWth CSPonD beam-down on-sun
generation concentrated solar power plants. Energ Conver Manage 2022;274:
demonstration plant. Energy Procedia 2015;75:2163–8.
116222.
[11] Kodama T, Gokon N, Matsubara K, Yoshida K, Koikari S, Nagase Y, et al. Flux
[25] Li P, Qian T, Li J, Lin H, Wang Y, Pei G, et al. Thermo-economic analysis of a novel
measurement of a new beam-down Solar concentrating system in Miyazaki for
partial cascade organic-steam rankine cycle. Energ Conver Manage 2023;283:
demonstration of thermochemical water splitting reactors. Energy Procedia 2014;
116941.
49:1990–8.
[27] Craig KJ, Moghimi MA, Rungasamy AE, Marsberg J, Meyer JP. Finite-volume ray
[12] Segal A. Solar energy at high temperatures; researches at the Weizmann Institute of
tracing using computational fluid dynamics in linear focus CSP applications. Appl
Science, Israel; 25 years of success. Renew Energy Environ Sustain 2016;1:1.
Energy 2016;183:241–56.
[13] Magaldi. Concentrated solar power and Thermal. https://www.magaldi.com/en/
[28] Soulayman S. Comments on solar azimuth angle. Renew Energy 2018;123:
applications/stemr-concentrated-solar-power-and-thermal. 2022 [accessed 3
294–300.
January 2024].
[29] Qiu Y, He YL, Li P, Du BC. A comprehensive model for analysis of real-time optical
[14] T.W. Bank. Solar resource data: Solargis. https://solargis.com/maps-and-gis-data/
performance of a solar power tower with a multi-tube cavity receiver. Appl Energy
download/world. 2020 [accessed 3 January 2024].
2017;185:589–603.
[15] N.R.E. Laboratory. Concentrating Solar Power Projects. https://solarpaces.nrel.
gov/project/yumen-xinneng-xinchen-50mw-beam-down. 2022 [accessed 3
January 2024].

24