Applied Thermal Engineering 118 (2017) 171–187

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Prospective fully-coupled multi-level analytical methodology

for concentrated solar power plants: General modelling
Lu Li a, Jie Sun b,⇑, Yinshi Li a,⇑
a
Key Laboratory of Thermo-Fluid Science and Engineering of MOE, School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, China
b
Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing 100190, China

h i g h l i g h t s g r a p h i c a l a b s t r a c t

A hybrid multi-dimensional model

(0-1-2-3 Model) for CSP plant is
proposed.

The model enables coupled system-
level, loop-level and component-level
analyses.

The model is advantageous in low
time-consuming and high
computational accuracy.

Solar-thermal-mechanical-electrical
energy conversions can be efficiently
investigated.

Optical-hydraulic-thermal-elastic
coupled issue can be
comprehensively addressed.

a r t i c l e i n f o a b s t r a c t

Article history: The complexity of the cascading solar-thermal-mechanical-electrical energy conversion in concentrated
Received 28 December 2016 solar power (CSP) plants urges to develop an accurate and fast analytical methodology for on-site use.
Revised 16 February 2017 Herein, we propose a novel fully-coupled multi-level analytical methodology, where multi-
Accepted 20 February 2017
dimensional model (0-1-2-3 Model) is developed to address the optical-hydraulic-thermal-elastic syner-
Available online 27 February 2017
gistic issue in CSP plants: (i) At system-level, the 0 Sub-Model reveals the heat-work transformation in
power block in the view of thermodynamics; (ii) At loop-level, the 1 Sub-Model uncovers the collection,
Keywords:
concentration and transformation of solar energy into the working fluid in the loop on account of 1D
Solar energy
Concentrated solar power (CSP)
thermo-hydraulics; (iii) The 2 Sub-Model, employing 2D finite volume method (FVM), figures out the
Parabolic trough collector (PTC) detailed circumferential temperature profile of receiver tubes in terms of composite heat transfer; (iv)
Direct steam generation (DSG) At component-level, the 3 Sub-Model, using 3D finite element method (FEM), brings insight into the
Multi-dimensional model nonuniform-temperature-induced deformation of receiver tubes focusing on the thermo-elastics. The
0-1-2-3 Model enables both system-level performance prediction and component-level targeting insight
in a remarkably high-efficient way with guaranteed accuracy. In comparison, the computational time of a
full 3D model is 15 times longer than that of the proposed 0-1-2-3 Model for a typical 4-m heat collection
element (HCE) and it goes up to 23 times longer for a 24-m loop section. To make the proposed model
easily understood, a CSP plant with parabolic trough collector (PTC) and direct steam generation (DSG)
technologies is applied. It was found that under the rated conditions, the energy and exergy efficiencies
of the plant are 18.89% and 20.26%, respectively.
Ó 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding authors.
E-mail addresses: sunjie@mail.etp.ac.cn (J. Sun), ysli@mail.xjtu.edu.cn (Y. Li).

http://dx.doi.org/10.1016/j.applthermaleng.2017.02.086
1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.
172 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187

Nomenclature

Abbreviations z z-direction in cylindrical coordinates;

CSP concentrated solar power
DSG direct steam generator system Greek letters
FEM finite element method a thermal diffusivity (m2 s1 ); thermal expansion coeffi-
FVM finite volume method cient (m K1 )
HCE heat collection element b volume expansion coefficient (K1 )
HTF heat transfer fluid d declination angle ( ); molecular diameter (m); principal
PTC parabolic trough collector stress (Pa)
PB power block e emissivity (–); strain (–)
SF solar field / latitude location ( )
c mass flow ratio of actual conditions and rated condi-
Symbols tions (–); ratio of specific heats (–); shear strain (–)
Aa aperture area (m2 ) gen energy efficiency (%)
BO the boiling number (–) gex exergy efficiency (%)
b interaction coefficient (–) gcoll collector efficiency (%)
1
cp isobaric thermal capacity (J kg K1 ) gfield solar field efficiency (%)
d hydraulic diameter (m) gpump pump efficiency (%)
DNI direct normal radiation (W m2 ) gturbine turbine efficiency (%)
E thermal flow rate (W); modifier factor for two phase k mean-free-path between collisions of a molecule (m)
flow heat transfer coefficient (coupled with S; E0 ; S0 ); l dynamic viscosity (Pa s); the Poisson’s ratio (–)
the Young’s modulus (MPa) m kinematic viscosity (m2 s1 )
Ex exergy flow rate (W) h incident angle ( ); circumferential direction in cylindri-
Fr the Froude number (–) cal coordinates
f friction factor (–) hrim rim angle of the collector ( )
1
G mass flux (kg s m2 ); shear modulus (MPa) htilt the tilt angle of PTC ( )
g gravity (9:81 ms2 ) hwet the wetting angle for stratified flow ( )
h heat transfer coefficient (W m2 K1 ); specific enthalpy hz zenith angle ( )
1
(J kg ) q density (kg m3)
IAM incidence angle modifier (–) r Stefan-Boltzmann constant (5:67  108 Wm2 K4 )
I exergy destruction rate (W) rS surface tension (mN m1 )
k conduction coefficient (W m2 K1 ) s shear stress (Pa)
1
L length (m) w physical exergy (J kg )
1
m_ mass flow rate in a loop (kg s ) x hour angle ( )

M _ total mass flow rate (kg s )

1
f vapor void fraction (–)
M relative molecular weight (–)
Nu the Nusselt number (–) Subscripts
p pressure (Pa) abs absorbed energy
Pr the Prandtl number (–) amb ambient
q heat flux (W m2 ) cond conduction heat transfer
Q heat flow (W) conv convection heat transfer
r radius (m); radial direction in cylindrical coordinates; g glass tube
R2 multiplier for pressure loss (–) in inlet of control volume
Ra the Rayleigh number (–) L liquid phase
Re the Reynolds number (–) v vapor phase
1
s specific entropy (J kg K1 ) tp two phase flow
T temperature (K) nb nucleate boiling
t temperature ( C) out outlet of control volume
u displacement (m); r receiver tube
W thermal power (W) rad radiation heat transfer
We net electric power (W) 0 heat transfer fluid
We the Weber number (–) 1 inner surface of receiver tube
x steam mass quality (–) 2 outer surface of receiver tube
X tt the Martinelli parameter (–) 3 inner surface of glass tube
y extraction ratio (–) 4 outer surface of glass tube

1. Introduction lection element (HCE) converts the concentrated radiation to

Concentrated solar power (CSP) is one of the most promising
large-scale sustainable energy technologies to feed the ever-
growing energy demand at near-zero environmental cost. When
operating a CSP plant, the optical concentrator, typically parabolic
trough collector (PTC) and heliostat, rises the normal solar radia-
tion to be a high energy-density state. Subsequently, the heat col-
thermal energy and transfers it into the heat transfer fluid (HTF)
in solar field (SF). Finally, the thermal energy is released into steam
for power cycling in power block (PB). Apparently, a CSP plant
involves the complicated solar-thermal-mechanical-electrical
energy conversion. Therefore, developing high-efficient numerical
models for composite heat transfer in SF and overall performance
is of great demand.
 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187
The heat transfer behaviors of the CSP plant were comprehen-
sively investigated by building one-dimensional (1D) models. For-
ristall [1] developed a 1D model for heat transfer analysis of linear
receiver implemented in Engineering Equation Solver (EES). Padilla
et al. [2] and Liang et al. [3] investigated the thermal performance
of PTC with their modified 1D models. Guo et al. [4,5] adopted a 1D
heat transfer model to achieve multi-parameter optimization
design and performance investigation of receiver tube of PTC. Kalo-
girou [6], Zaversky et al. [7] and Manzolini et al. [8,9] also pre-
sented the 1D thermo-hydraulic models to conduct thermal
analysis.
The solar-thermal energy conversions of the CSP plant were sys-
tematically analyzed by developing the two-/three-dimensional
(2D/3D) models. He et al. [10,11] proposed a Monte Carlo Ray Trace
(MCRT) and finite volume method (FVM) coupled simulation
method for the energy conversion analysis of PTC. Afterwards,
other researchers developed similar methods to investigate the
heat transfer, such as Hachicha et al. [12], Qiu et al. [13] and Wu
et al. [14]. Wang et al. [15] constructed a 2D model for the
radiation-convection composite heat transfer using FVM and stud-
ied the asymmetric characteristics, including the incident angle
and nonuniform heat flux. Eck and Steinmann [16] developed a
2D model to investigate the thermo-hydraulic behavior in direct
steam generation (DSG) field and presented a simplified model to
acquire the temperature distribution of receiver tube. Wang et al.
[17] proposed the solar trace method and finite element method
(FEM) to solve the conjugate heat transfer and thermal stress prob-
lem of receiver tube. Khanna et al. [18] derived an explicit analyt-
ical expression to calculate the temperature distribution of
receiver tube and a bimetallic tube was proposed to reduce the cir-
cumferential temperature gradient. Wang et al. [19] studied the
influence of glass tube on heat flux distribution and proposed
elliptic-circular cross-section envelope to reduce the circumferen-
tial flux gradient. Wirz et al. [20] proposed a pseudo 3D optical and
thermal numerical model and investigate the temperature distri-
bution of receiver tubes. Marugán-Cruz et al. [21] investigated
the thermal analysis of the tubes under nonuniform heat flux
and gave the relationship between the Biot number and the accu-
racy of 1D model. Lobón et al. [22] implemented STAR-CCM+ to
simulate the multiphase behavior of a PTC-CSP plant with DSG.
In addition, Serrano-Aguilera et al. [23] built a 3D thermal model
for superheating stage in PTC with DSG to acquire the 3D temper-
ature field of receiver tube and glass tube.
The overall performances of the CSP plant were also widely
studied by modelling the zero-dimensional (0D) thermodynamic
cycle. Zarza et al. [24] designed a 5-MWe CSP plant with DSG in
the INDITEP project and exhibited good performance. According
to SEGS VI plant, Patnode [25] adopted a thermal loss model for
the PTC using Therminol VP-1 as HTF and a linear regression Rank-
ine cycle model to assess the overall performance. Montes et al.
[26,27] studied a 50-MWe CSP plant with DSG, thermal storage
and auxiliary natural gas-boiler and evaluated the influence of
the solar multiple on the annual performance. Biencinto et al.
[28] presented a quasi-dynamic simulation model for industrial
CSP plant with DSG using TRNSYS to make long-term power anal-
yses. Manzolini et al. [8,9] developed the 0D thermodynamic cycle
model to estimate performance of different CSP plants, as well as
the sizing and investment costs. The performance investigation
was also carried out by Al-Sulaiman et al. [29] and Desai et al.
[30] based on the 0D thermodynamic cycle model.
Literature review indicates that previous modellings are mostly
based on a uni-dimensional and isolated view, and only focus on
one single aspect of CSP plant. For example, an excessively detailed
(3D) model of optical-thermal energy conversion provides
component-level insights but makes no contribution to the
system-level understanding simultaneously. On the contrary, an
173
excessively simplified (0D) model for the system analysis abso-
lutely loses the local details. It is noted that the component-level
details and the system-level performances are inherently coupled
and the system performance could be significantly affected by mal-
function of a single component. Therefore, a comprehensive ana-
lytical methodology that enables both system-level performance
prediction and component-level targeting insight with a decent
rate and accuracy is urgently needed. To this end, the idea of
fully-coupled multi-level analytical methodology for CSP plants is
originally proposed in the present work based on the novel hybrid
multi-dimensional model (0-1-2-3 Model). The 0-1-2-3 Model is
capable of completing the solar-thermal-mechanical-electrical
energy conversions at low time-consuming cost and high accuracy.
Simultaneously, the optical-hydraulic-thermal-elastic synergistic
issues in CSP plants are specially concerned. The temperature
and stress fields of any risky section in a loop can be automatically
detected and acquired. Moreover, the effect of the deformation of
the receiver tubes can be fed back to the system performance for
correction. A DSG system with recirculation mode has been
adopted for demonstration and the corresponding results and val-
idations have been conducted. Compared with existing models, the
0-1-2-3 Model has demonstrated distinct merits.
2. Methodology
The proposed fully-coupled multi-level analytical methodology
for CSP plants technically benefits from the proposed 0-1-2-3
Model, which can be broken down into four Sub-Models, namely
0 (0D), 1 (1D), 2 (2D), and 3 (3D) Sub-Models, as shown in Fig. 1.
The 0 Sub-Model is applied to PB involving the transformations
from the thermal energy of steam to work by the steam turbine
and to the electric energy by the generator. The 1 Sub-Model is
employed to calculate the thermo-hydraulics in the loops of SF.
The 2 Sub-Model, benefitting FVM, aims to work out the circumfer-
ential temperature distribution of the receiver tube of HCE. The 3
Sub-Model, adopting FEM, undertakes the thermal-stress-strain
analysis of HCEs. For each run, the 0 Sub-Model is initialized with
reasonable values based on the configuration of PB and provides
inputs for the 1 Sub-Model of SF. The 1 Sub-Model, with the inlet
conditions from the 0 Sub-Model, solves the 1D thermo-
hydraulics along the loop, of which the results at outlet are given
as the inputs of PB. The 0-1 Sub-Model forms an outer iterative
loop aiming on the thermodynamics of the CSP plant, e.g. the
energy and exergy efficiencies and the irreversible losses. When
the convergent results are acquired, the overall performance of
the CSP plant is readily to be evaluated. On the other hand, the flow
pattern in the evaporation stage and the local temperature at the
outlet of superheating stage are monitored simultaneously when
the 1 Sub-Model is working. Once the risky conditions are
detected, i.e. the stratified flow emerging in the evaporation stage
or the local temperature beyond the threshold in the superheating
stage, the 2 Sub-Model is immediately awakened to calculate the
detailed 2D temperature distribution within the risky sections.
Note that the 2 Sub-Model benefits from the initial temperature
distribution provided by the 1 Sub-Model, therefore a finer conver-
gent 2D (circumferential and axial directions) results can be readily
achieved. This 2D temperature distribution is automatically
checked whether the maximal circumferential temperature differ-
ence is beyond the threshold for potential thermal-stress-induced
deformation problem of HCE. If yes, the 2D temperature distribu-
tion of the risky section is directly transferred into the 3 Sub-
Model as thermal boundaries to further solve the thermo-
elastics. The 2-3 Sub-Model forms an inner iterative loop aiming
at the thermos-elastics of the HCEs, e.g. thermal stress and strain.
When the convergent results are acquired, the thermal-stress-
174 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187

Fig. 1. Schematic of hybrid multi-dimensional model.

 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187
induced deformation of the HCEs in loop is readily to be evaluated.
Based on the deformation results, a reconstructed solar radiation
distribution over the deformed receiver tube can be derived. With
the optical correction by the reconstructed solar radiation distribu-
tion, the 0-1 Sub-Model is recalled and the overall performance of
the CSP plant is therefore corrected.
The flexible collaboration between Sub-Models shows promi-
nent advantages and contributes to carrying out researches from
multiple perspectives. By properly organizing the Sub-Models,
not only the system-level evaluation but also the loop-level and
even the component-level insights of the CSP plant are readily
achievable. The 0-1 Sub-Model presents the results in the
system-level and loop-level, e.g. the energy delivery and conver-
sion, and the thermo-hydraulics along the loop. The 2-3 Sub-
Model provides a detailed description of receiver tube in the
component-level. Moreover, the system-level analysis can receive
and handle the feedback information from the component-level
analysis. Then the influence on the total system arising from the
abnormal performance of components, such as the deformation
of the receiver tube, is presented clearly. Also, the arrangement
occupies less time in computation while guaranteeing high accu-
racy. The superiority stands out in the 2-3 Sub-Model. Acquiring
the tube temperature field mainly aims to serve the thermo-
elastic analysis. Making a thorough 3D thermal analysis is of course
able to gain more accurate results. However, note that even the full
3D modelling of only one complete loop (hundreds of meters long)
is extremely time-consuming and memory-intensive, demanding
for high-performance computing (HPC).
For PTC, the solar radiation concentrated by the parabolic
reflecting mirrors penetrates through the glass tube, reaches the
outer surface of receiver tube and is absorbed and converted into
thermal energy transferred to the HTF inside the receiver tube
via conduction and convection. Meantime, a part of the thermal
energy is dissipated as heat loss through the convection and radi-
ation from the outer surface of glass tube to the ambient and sky.
In particular, phase-change has to be taken into serious considera-
tion when dealing with the flow boiling problem inside the recei-
ver tube, where water/steam serves as both HTF and working
fluid in the DSG technology. When operating in the recirculation
mode, which has been proven the optimal for DSG [31], it is impor-
tant to regulate the main flow rate to maintain a certain steam
mass quality at the water-steam separator, and the injection flow
rate to keep the desired outlet temperature. The layout of a typical
CSP plant is shown in Fig. 1.
3. Modelling
In the present work, the 0-1-2-3 Model is specially developed to
tackle the multi-level problems in the PTC-CSP plant with DSG. The
flowchart of code is shown in Fig. 2. Firstly, the 0 Sub-Model is ini-
tialized to offer the inlet conditions for the 1 Sub-Model of SF.
Then, combing the meteorological data, the 1 Sub-Model is
launched to calculate the solar-thermal energy conversion. In the
1 Sub-Model, the typical three stages of thermo-hydraulics, i.e.
preheating, evaporation and superheating, are considered. Accord-
ingly, the states of subcooling water, water-steam saturation and
superheating steam can be distinguished by the corresponding
enthalpy. After the 1 Sub-Model converges, the steam information
at the outlet of loop is determined and transferred back to the 0
Sub-Model for heat-work transformation calculation. On the other
hand, the 2-3 Sub-Model performs conditionally when the risky
conditions are detected. Once being called, the 2 Sub-Model starts
to calculate the 2D temperature distribution of receiver tube ben-
efitting the initial values from the 1 Sub-Model. Afterwards, the
temperature distribution is delivered to the 3 Sub-Model and the
175
thermal-stress-strain calculation is executed to quantify the axial
deflection of the receiver tube. With the axial deflection, the corre-
sponding reduction in the solar field efficiency, i.e. the optical cor-
rection, is calculated and the thermal energy collected from the
solar radiation is therefore updated. Finally, with the aid of the
updated solar field efficiency, the 0-1 Sub-Model is recalled to
recalculate the thermodynamic cycle under the updated operating
conditions. Note that if the axial deflection is beyond the safety
limit, meaning that the glass tube takes the risk of being broken,
the code immediately gives a warning. Apart from the 3 Sub-
Model implemented in ANSYS through external call, the rest of
the code is wholly self-developed in MATLAB. The thermophysical
properties of water/steam are acquired according to the IAPWS-
IF97 formula.
3.1. The 0 Sub-Model
A detailed mathematical model for PB is burdensome and off
the core of the present work, therefore a reasonably simplified
model is built. On the whole, the 0 Sub-Model determines the rela-
tionship between the thermodynamic states of main steam and the
output work regardless of delayed effect of all other units. The
INDITEP project [24] has been referred to in the configuration for
SF (see Table 1). More details of the plant can be found in Table 2.
The temperature-entropy (T-s) diagram for the thermodynamic
cycle is shown in Fig. 3, where 4-5 and 5-6-7 correspond to the
processes in SF and steam turbine, respectively, and 6 s and 7 s
are isentropic states. The pressure and temperature of relevant
state points are also given in Table 3.
The main components of PB include the steam turbine, con-
denser, pumps, deaerator and electric generator. The correspond-
ing models are given below.
(1) Steam turbine
The mechanical work generated by the steam turbine is related
to the enthalpy difference and the mass flow rate:
_ in  hout Þ
W turbine ¼ mðh ð1Þ
The outlet enthalpy influenced by the stage efficiency and the isen-
tropic efficiency is expressed as follows:
hin  hout
gturbine ¼ ð2Þ
hin  hout;s
where hin is the inlet enthalpy, hout is the outlet enthalpy, hout;s is the
outlet enthalpy in response to the isentropic process, m _ is the mass
flow rate. The stage efficiency of turbine varies with the load. There-
fore, the practical value is modified as [25]:
gturbine ¼ grated ð0:809 þ 0:409c  0:218c2 Þ ð3Þ
where c ¼ m= _ m_ rated , m
_ rated is the mass flow rate at the rated
condition.
(2) Pump
Pump performance is modeled similar to the turbine, and the
isentropic efficiency can be expressed as:
hin  hout;s
gpump ¼ ð4Þ
hin  hout
where the pump efficiency is assumed constant due to the negligi-
ble variation. The work consumed by the pump is calculated as:
_ out  hin Þ
W pump ¼ mðh ð5Þ
(3) Condenser and deaerator
It is assumed that exhaust steam from turbine is cooled to sat-
uration state by the condenser. The detailed physical model is
176 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187

Fig. 2. Flowchart of code.

complicated in deaerator where the extracted stream is mixed flow rate at the outlet of deaerator is the sum of the main stream
with feedwater. For simplicity, the state of mixture at outlet of and extraction rates, and the energy equation in deaerator is
the deaerator can be assumed as saturation. Therefore, the mass expressed as:
 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187 177

X X X
Table 1 min hin  mout hout ¼ W  Qi ð7Þ
Configurations of INDITEP project [24]. in out i

Quantity (unit) Value X X X

Overall length of a single collector (m) 98.5
min win  mout wout ¼ W þ I  ð1  T 0 =T i ÞQ i ð8Þ
in out i
Number of parabolic-trough modules per collector (–) 8
Gross length of every module (m) 12.27
where I ¼ T 0 sgen and w ¼ h  h0  T 0 ðs  s0 Þ. I is exergy destruction
Aperture (m) 5.76
Outer diameter of receiver tube (m) 0.07 and w is physical exergy for a given state.
Inner diameter of receiver tube (m) 0.055 For PB, the total received energy and exergy are:
Outer diameter of glass tube (m) 0.125
_ 5  h4 Þ; DEx;block ¼ Mðw
DEblock ¼ Mðh _
Inner diameter of glass tube (m) 0.119 5  w4 Þ ð9Þ
Direct solar irradiation (Wm2 ) 875
Geographical longitude of the site (°) 5.97 W
The net electric power generated from the generator is expressed
Geographical latitude of the site (°) 37.4 N as:
Inlet water temperature (°C) 153 X
Inlet water pressure (bar) 80 W e ¼ gg Wi ð10Þ
Outlet steam temperature (°C) 410 i
Outlet steam pressure (bar) 69.9
gg ¼ 0:9 þ 0:258load  0:3load2 þ 0:12load3 ð11Þ

Table 2 where load is the ratio of the turbine power over the rated power
Operational conditions at design-point. [25]. Note that the pump consumes power so the negative sign is
used to distinguish from the output work.
Quantity (unit) Value
2
The total received energy and solar radiation exergy by SF are
Rated DNI (W m ) 800
calculated in Eq. (12), as [33]:
Inlet temperature (°C) 420
X
Inlet pressure (bar) 80
Esolar ¼ Q abs ; Ex;solar
Rated mass flow rate (kg s1) 65 "
Single loop outlet flow rate (kg s1) 1.13  4  #
1 T amb 4 T amb
Number of total loops (–) 58 ¼ 1þ   DNI  A ð12Þ
Rated electric power (MWe) 50 3 T sun 3 T sun
Turbine efficiency (%) 72
Pump efficiency (%) 60 T amb is the ambient temperature, T sun is the Sun’s temperature,
Back pressure (bar) 0.1 5777 K, A is the total aperture area of SF.The energy and exergy
Ambient temperature (°C) 20 acquired by HTF are as follows:
Wind speed (m s1) 1.2 X X
Collector field inlet temperature (°C) 150 DEHTF ¼ m_ out hout  m_ in hin ; DEx;HTF
Collector field inlet pressure (bar) 90 out in
Geographical longitude of the site (°) 5.97 W X X
Geographical latitude of the site (°) 37.4 N ¼ m_ out wout  m_ in win ð13Þ
out in

where the inlet and outlet include the re-injection ports and the
separation ports, respectively.
Energy and exergy efficiencies of PB are calculated as:
5
We We
gen;block ¼ ;g ¼ ð14Þ
DEblock ex;block DEx;block
Energy and exergy efficiencies of SF are:
T

DEHTF DEx;HTF
gen;field ¼ ;g ¼ ð15Þ
Esolar ex;field Ex;solar
6
4 Energy and exergy efficiencies for the CSP plant are:
3 We We
6s
gen;overall ¼ ;g ¼ ð16Þ
2 Esolar ex;overall Ex;solar
1
7s 7 3.2. The 1 Sub-Model
S
In SF, water is heated by the concentrated radiation energy to a
Fig. 3. Temperature-entropy diagram for the power cycle; 4-5 occurs in SF; 5-6-7-
desired thermodynamic state. The basic solar-thermal conversion
1-2-3-4 occurs in PB.
is that the solar radiation penetrating the glass tube reaches the
outer receiver tube covered with the selective absorption coating.
Then the thermal energy is transferred to the HTF by conduction
hout ¼ yhextract þ ð1  yÞhstream ð6Þ
and convection. The energy losses mainly come from the radiation
where hout is the outlet enthalpy, hstream is the inlet enthalpy, hextract and convection between the outer glass tube and the surroundings.
is the extraction stream enthalpy, and y is the extraction ratio. Note Non-vacuum condition in the annular space between the receiver
that the extraction ratio and extraction pressure are calculated by tube and the glass tube has been taken into consideration. The 1
iteration. Sub-Model has been developed to acquire the thermo-hydraulic
(4) Energy and exergy parameters along the flow direction efficiently. The process relates
Energy and exergy balances for a control volume are expressed to four contact surfaces, which can be simplified to six nodes (see
in Eqs. (7) and (8), respectively, as [32]. Fig. 4). In Fig. 4, 0–4 stand for the HTF, inner surface of receiver
178 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187

Table 3
Thermodynamic properties of state points in rated cycle.

State point Pressure/bar Temperature/°C 1 1

Enthalpy/kJ kg Mass flow rate/kg s
1 0.1 45.81 191.8 54.4
2 4.47 45.9 192.5 54.4
3 4.47 147.64 622.1 11
4 90 150 637.6 65.4
5 80 420 3194.2 65.4
6s 4.47 147.64 2569.6 65.4
6 4.47 148.27 2744.5 65.4
7s 0.1 45.81 2173.4 54.4
7 0.1 45.81 2333.9 54.4

transfer fluid. h is the convective coefficient. T is the absolute tem-

perature while t is in the unit of Celsius. Also, the adjacent units
meet the relation, as:

Q gain
tout ¼ t in þ ð26Þ
cp m_

where Q gain is the heat gain of HTF and cp is the specific heat capac-
ity of HTF.
(1) Absorbed solar energy
When HCE is rotated about a horizontally south-north tracking
axis, the solar radiation absorbed by HCE can be calculated as:
Q abs ¼ DNI  cos h  IAM  gfield  gcoll  Aa ð27Þ
where DNI is the direct normal irradiation, h is the incidence angle,
IAM is the incident angle modifier, gfield is the field efficiency, gcoll is
the collector efficiency and Aa is the aperture area. The incidence
angle is defined as [34]:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
cos h ¼ cos2 hz þ cos2 d sin x ð28Þ

cos hz ¼ cos d cos / cos x þ sin d sin / ð29Þ

Fig. 4. Optical-thermal conversion process with the 1 Sub-Model of HCE. 0-Water/
steam; 1-inner surface of receiver tube; 2-outer surface of receiver tube; 3-inner where hz is the zenith angle, d is the declination angle, x is the hour
surface of glass tube; 4-outer surface of glass tube. angle and / is the latitude location. IAM that is given according to
experimental data stands for the losses due to additional reflection
tube, outer surface of receiver tube, inner surface of glass tube and and absorption by the glass tube. In the present work, the LS-2 col-
outer surface of glass tube, respectively. The rest is related to the lector is chosen and IAM is calculated as [35]:
surroundings. The energy balances are as follows.
q10;conv ¼ q21;cond ð17Þ IAM ¼ 1 þ 8:84E4 h= cos h  5:369E5 h2 = cos h ð30Þ
In addition, to quantify the geometry and sun-tracking of PTC, the
q21;cond ¼ qabs  q23;conv  q23;rad ð18Þ angular parameters are required. The rim angle, hrim , is the angle
between the symmetric line and the rim of the reflecting mirror
q34;cond ¼ q23;conv þ q23;rad ð19Þ (see Fig. 5). Due to the existence of hrim , the concentrated solar radi-
ation is not uniformly distributed around the receiver tube in the
q34;cond ¼ qsky;rad þ qamb;conv ð20Þ circumferential direction. The tilt angle, htilt , is the angle between
the normal direction of reflecting mirror and the gravitational direc-
with tion, which exerts significant impact on the two-phase flow pattern
q10;conv ¼ h10;conv ðt 1  t 0 Þ ð21Þ inside the receiver tube. htilt is calculated as [34]:
 
cos hz
eg rðT 42  T 43Þ htilt ¼ arccos ð31Þ
q23;rad ¼ ð22Þ cos h
1
er þ 1
eg 1 (2) Heat transfer in receiver tube
As for single-phase flow, the heat transfer coefficient is deter-
qsky;rad ¼ eg rðT 44  T 4sky Þ ð23Þ mined as:
hconv ¼ kdNu ð32Þ
qamb;conv ¼ hamb ðt 4  t atm Þ ð24Þ
with Nu ¼ 4:36 for laminar flow. If the flow is turbulent, Nu could be
kDt calculated according to the Gnielinski equation [36], as:
qcond ¼ ð25Þ
r lnðr outer =r inner Þ  0:11
ðf =8ÞðRed  1000ÞPr1 Pr1
Nu ¼ pffiffiffiffiffiffiffiffi   ð33Þ
where k is the conduction coefficient, e is the emissivity, r is the Pr2
1 þ 12:7 f =8 Pr2=3 1  1
Stefan-Boltzmann constant, t 0 is the average temperature of heat
 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187 179

To numerically describe the stratified flow, the wetting angle,

hwet , standing for liquid-phase proportion, is employed (see
Fig. 7). The angle is determined from the implicit equations [42] as:

AL ¼ 0:5r 21 ½hwet  sin hwet ; AL ¼ Að1  1Þ ð42Þ

where AL is the cross-sectional area occupied by the liquid-phase
and 1 is the vapor void fraction [43], as:
(    0:25 )1
x x 1x 1:18ð1  xÞ g rs ðqL  qV Þ
1¼ ½1 þ 0:12ð1  xÞ þ þ
qV qV qL G q2L
ð43Þ

where rs is the surface tension.

When stratified flow comes into being, the uneven thermal load
on the receiver tube will be enlarged because of the pronounced
difference in heat transfer coefficients between liquid and gas
phases. To clarify the process, some typical cross-sectional profiles
are shown in Fig. 7. Considering the uneven radiation distribution,
the circumference region can be split up into four regions: inner
Fig. 5. Cross-sectional view of PTC. wetting with outer radiation, inner wetting with outer radiation-
free, inner non-wetting with outer radiation and inner non-
wetting with outer radiation-free. Among them, the third case is
1 the worst occasion.
f ¼ 2
ð34Þ
ð1:82log10 Red  1:64Þ
(3) Heat transfer in annular space
where 0:5 < Pr 1 < 200 and 2300 6 Red 6 5  106 .
In DSG system, considering the gravitation, the flow patterns The convection in the annular space which is associated with
show asymmetry in horizontal tubes. Along the flow direction, the inner pressure has been adopted. If the pressure is less than
there exist various flow patterns, e.g. bubbly flow, plug flow, slug 133 Pa, the process is free-molecular convection, otherwise, it is
flow, wavy flow, stratified flow and annular flow. Among them, natural convection [1]. For free-molecular convection:
annular flow is expected while stratified flow is unwanted due to
potential thermal stress issues [37]. Flow boiling heat transfer q23cov ¼ pd2 h23;cov ðt 2  t3 Þ ð44Þ
inside horizontal tubes is determined by the flow pattern which
kv;0
can be distinguished by the Taitel-Dukler flow regime map [38]. h23;cov ¼     ð45Þ
r3
In evaporation stage, the flow boiling heat transfer process is the r 2 ln r2
þ bk rr23 þ 1
combined mechanism of convective boiling and nucleate boiling.
The local two-phase boiling coefficient is obtained from the 9c  5
Gungor-Winterton correlation [39,40], as: b¼ ð46Þ
2ðc þ 1Þ
htp ¼ Ehconv þ Shnb ð35Þ
1:7484Eð22ÞT ave
where convective coefficient hcnov is calculated from the Dittus- k¼ ð47Þ
Pv d2
Boelter correlation, as:

hconv ¼ 0:023Re0:8 0:4 T2 þ T3

L Pr L kL =d ð36Þ T ave ¼ ð48Þ
2
and the nucleate boiling coefficient hnb is calculated from the
where kv ;0 is the conductivity of annular gas in standard conditions,
Cooper correlation [41], as:
d is the molecular diameter of annulus gas, k is the mean-free-path
0:55
hnb ¼ 55p0:12
r ðlog10 pr Þ M 0:5 q0:67 ð37Þ
with ReL using the local liquid fraction of the flow GL , pr ¼ p=pc , M is 5
the molecular weight and pc is the critical pressure of water. When Gaussian distribution

the flow is unstratified, E0 and S0 are expressed as: Rectangular distribution

4
-2

E0 ¼ 1 þ 24; 000Bo1:16 þ 1:37X 0:86

Solar radiation flux / Wcm

tt ð38Þ

 1 3
S0 ¼ 1 þ 0:00000115E20 Re1:17
L ð39Þ Practical distribution

where Bo is the boiling number and X tt is the Martinelli parameter. 2

 0:9  0:5  0:1
q_ 1x qV lL
Bo ¼ ; X tt ¼ ð40Þ
GDhV x qL lV 1

For stratified flow, E0 and S0 must be multiplied by the correction

factors, as: 0
0 30 60 90 120 150 180 210 240 270 300 330 360
0
E ¼ Fr 0:12Fr
L
L
; S0 ¼ Fr0:5
L ð41Þ Circumferential angle / o

where Fr L is the Froude number of liquid, FrL ¼ G2L =ð L gd1 Þ.

2
q Fig. 6. Distributions of concentrated solar radiation flux.
180 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187

Fig. 7. Schematics of stratified flow coupling with uneven solar radiation distribution (red). (a) and (b): htilt ¼ 0; (c) and (d): htilt –0.

between collisions of a molecule, c is the ratio of specific heats of Pressure drop along the flow direction is acquired according to
the annular gas and b is the interaction coefficient. the Blasius equation [45], the pressure loss for single-phase turbu-
For natural convection [44]: lent flow is determined by the flowing equations:

0:386kave ðt2  t 3 ÞðPrave Ra2 =ð0:861 þ Prave ÞÞ0:25 2f DLG2

q23cov ¼ ð49Þ Dps ¼ ð55Þ
r2 ð1 þ ðr 2 =r 3 Þ0:6 Þ
1:25
qs d1

gbðt 2  t 3 Þr32 f ¼ 0:079Re0:25

d1 ð56Þ
Ra2 ¼ ð50Þ
am where Dps is the pressure drop for pipe with DL length, G is the
where Ra2 is the Rayleigh number with characteristic length r 2 and single-phase mass flux, f is the fraction factor. While for the two-
a is the thermal diffusivity. For ideal gas, b ¼ 1=T ave . phase flow, the DISS project has verified the Friedel correlations
(4) Heat transfer outside glass tube [16,46], as:
The outer surface of glass tube is connecting with the surround-
Dptp ¼ R2 Dps ð57Þ
ings and when natural convection plays a part around a horizontal
cylinder, the Churchill and Chu correlations [36] are used as: where Dps is calculated according to Eq. (55) assuming only liquid-
( )2 phase fluid occupies the flow channel. The multiplier R2 is calcu-
1=6
0:387Ra
Nu ¼ 0:6 þ 8=27
ð51Þ lated with the next formula, including some dependent dimension-
½1 þ ð0:559=PrÞ9=16  less factors, as:

 4  T amb Þd 3 qL f V 3:24x0:78 ð1  xÞ0:224 H

g bðT R2 ¼ ð1  xÞ2 þ x2 þ ð58Þ
Ra ¼ 4
ð52Þ qV f L Fr 0:045 We0:035
m
a  H L

 0:91  0:19  
 ¼ 1 ; T ¼ ðT 4 þ T amb Þ qL lV l 0:7
b ð53Þ H¼ 1 V ð59Þ
T 2 qV lL lL
 and m
where Pr, a  are the Prandtl number, thermal diffusivity and

kinematic viscosity of air respectively, all of which are based on T. G2 d1
WeL ¼ ð60Þ
When forced convection performs, Nu is expressed as: rs qave
Nu ¼ CRend Pr1=3 ð54Þ  1
x 1x
qave ¼ þ ð61Þ
where Red is based on d4 , 0:4 < Red < 4  105 , the parameters C and qV qL
n are dependent on Red [45]. where WeL is the Weber number, qave is the homogeneous density,
(5) Pressure loss Frave is the Froude number based on qave . For laminar flow
 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187 181

(Re < 1055), f ¼ 16=Re while for turbulent flow (Re > 1055), @dr
@r
þ 1r @@h
shr
þ @@z
szr
þ dr d
r
h
¼0
2
f ¼ ½1:73718 logð1:964 logðReÞ3:8215
Re
Þ , where Re is calculated based @ srh
@r
þ 1r @d@h
h
þ @@z
szh
þ 2 srrh ¼ 0 ð67Þ
on the discrete phases respectively. @ srz 1 @ shz
þ r @h þ @z @dh
þ srz
¼0
@r r

3.3. The 2 Sub-Model According to the equivalence theorem of shear stress,

shr ¼ srh ; srz ¼ szr ; szh ¼ shz ð68Þ
The 2 Sub-Model aims to acquire the tube wall temperature dis-
tribution. The model is implemented with FVM and the following As for thermo-elastic problems, the constitutive equations for a
makes particular description. homogeneous, isotropic body is expressed in cylindrical coordinate
The control equation for the conduction without source terms system as:
in cylindrical coordinate system is reduced to the following er ¼ 1E ½dr  lðdh þ dz Þ þ aDt
equations:
   
eh ¼ 1E ½dh  lðdz þ dr Þ þ aDt ð69Þ
1 @ @t 1 @ @t ez ¼ 1E ½dz  lðdr þ dh Þ þ aDt
kr þ 2 k ¼0 ð62Þ
r @r @r r @h @h
srh shz szr
The compound heat transfer is involved in several physical surfaces. crh ¼ ; chz ¼ ; czr ¼ ð70Þ
G G G
The schematic of HCE and the configuration of grid are shown in
Fig. 8. The boundary conditions are given as follows: where E is the Young’s modulus, l is the Poisson’s ratio, a is the
  coefficient of thermal expansion and G is the shear modulus,
@t G ¼ 2ð1þ
E
kr ¼ q10;conv ð63Þ lÞ.
@n 1 The strain-displacement relations are as follows:
  @ur ur 1 @uh @uz
@t er ¼ ; eh ¼ þ ; ez ¼ ð71Þ
kr ¼ qabs  q23;conv  q23;rad ð64Þ @r r r @h @z
@n 2

  crh ¼ 1r @u@hr þ @u@rh  urh

@t
kg ¼ q23;conv þ q23;rad ð65Þ
@n 3 chz ¼ 1r @u@hz þ @u@zh ð72Þ
  czr ¼ @u@zr þ @u@rz
@t
kg ¼ qair;conv þ qsky;rad ð66Þ Considering that both ends of a receiver tube have the corrugated
@n 4
pipe connection joint, the end restrains are equivalently free.
Different radiation flux distribution over the radiation region can be Stress evaluation is based on the von Mises equivalent stress
adopted (see Fig. 6). However, other relevant correlations keep con- which is calculated as:
sistent with the 1 Sub-Model. sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðdr  dh Þ2 þ ðdh  dz Þ2 þ ðdz  dr Þ2
de ¼ ð73Þ
3.4. The 3 Sub-Model 2

For 3D axisymmetric elasticity problems, regardless of body

force, the equilibrium equations in a cylindrical coordinate system
can be expressed as [47]:
Fig. 8. Schematic of HCE and configuration of grid for the 2 Sub-Model.
3.5. Data delivery between Sub-Models
From the stress-strain analysis for a specific receiver tube to the
electric power output of the overall CSP plant, the 0-1-2-3 Model is
efficient in handling the solar-thermal-mechanical-electrical
energy conversions at multiple levels. The crux of the matter
resides in collaboration and information transfer.
At the design point, the 0-1 Sub-Model collectively operates to
accomplish the configurations of SF and PB. On-the-fly, the outlet
data from the 1 Sub-Model, e.g. mass flow rate and pressure, are
transferred to the 0 Sub-Model of PB (1 ? 0). Meanwhile, the out-
let data from PB are also returned to the 1 Sub-Model as inlet con-
ditions (0 ? 1). A more time-saving way of execution for 2 Sub-
Model is to take the results of the 1 Sub-Model as the initial value
(1 ? 2). The 2D temperature distribution from 2 Sub-Model offers
to the 3 Sub-Model as the thermal boundaries (2 ? 3). An essential
point lies in the consistency of grid structure between the 2 and 3
Sub-Models. In this work, for a single typical HCE, the axial length
is divided into 8 segments and the circumferential length into 90
cells in both Sub-Models. The node information in the 2 Sub-
Model precisely satisfies the demand of the 3 Sub-Model. The
SOLID186 Homogeneous Structural Solid Element has been
adopted for the 3 Sub-Model. All the manipulations are actualized
carefully with the ANSYS Parametric Design Language. The 3
Sub-Model benefits FEM solver with distinct advantages in
thermo-elastic analysis. In addition, the deformation of receiver
tubes probably leads to decrease of the collector efficiency.
Therefore, the 3D bending information should be fed back to the
182 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187

Table 4 prominent. This strategy of targeting insights helps to operate

Configurations of DISS project.
Quantity (unit)
Aperture (m)
Focal length (m)
Outer diameter of receiver tube (m)
Inner diameter of receiver tube (m)
Outer diameter of glass tube (m)
Inner diameter of glass tube (m)
Geographical longitude of the site (°)
Geographical latitude of the site (°)
Optical efficiency (%)
Table 5
Experimental data from Lobón et al. [22].
Label pin /MPa t in / C
Case 4 6.23 ± 0.06 235 ± 2
Case 5 10.19 ± 0.06 237 ± 2
1 Sub-Model of SF to quantify the solar radiation loss and further to
the 0-1 Sub-Model to update the overall efficiency of CSP plant
considering the influence of deformation of receiver tubes (3 ?
1 & 0).
The data delivery between the 0 and 1 Sub-Models forms the
basically outer iterative loop, which is carried out unconditionally
at system-level and loop-level. While the 1 ? 2 ? 3 data deliveries
are carried out conditionally at component-level. Once the defor-
mation of the receiver tubes is considered, 3 ? 1 & 0 together with
1 ? 2 ? 3 data deliveries immediately form the inner iterative
loop.
Notably, the Sub-Models can be flexibly implemented with a
different strategy, where both the 0-1 and 2-3 Sub-Models are car-
ried out intentionally for targeting insights. This feature enables
the proposed 0-1-2-3 Model to be a diagnosis-assistant tool that
can be used to detect stress and deformation of specific receiver
tubes or segments of loop. The significant advantage is particularly
applicable in DSG plants where the stress-strain problem is more
more safely.
Value
5.57 [31] 4. Results and discussion
1.71 [31]
0.07 [48]
0.05 [48]
4.1. The 1 Sub-Model
0.12 [48]
0.115 [48] The realistic case from DISS project is compared with the 1 Sub-
2.36 W [31] Model in once-through mode, consisting of 9 collectors in 50-m
37.1 N [31]
and 2 collectors in 25-m. The collector configuration information
0.73 [31]
and the selected case data are listed in Tables 4 and 5, respectively.
According to M. Eck [31], the incident angle modifier of the
used 25- and 50-m LS-3 collectors has been updated with
IAM25 ¼ 1  0:00362h  1:32337  104 h2 and IAM 50 ¼ 1
0:00188h  1:49206  104 h2 . For case 4, the maximum deviations
Mass flow/kg s
1
DNI/W m2 of temperature and pressure are 4.8% and 0.36% (Fig. 9a),
0.5 ± 0.05 850 ± 10 respectively. As for case 5, the values are 6.6% and 0.35%
0.59 ± 0.05 960 ± 10 (Fig. 9b), respectively. All the deviations are still acceptable.
The 1 Sub-Model is also validated by comparing the present
results with experimental data from the INDITEP project [24].
Detailed configuration can be found in Table 1. The calculation con-
ditions include DNI, 875 Wm2 , the steam mass quality at separa-
tor, 0.8, and the outlet temperature, 410 °C. The comparisons are
given in Fig. 10. The results include pressure, temperature,
enthalpy, steam mass quality and mass flow rate. Compared with
Zarza et al.’s results, good agreement is achieved for the present
1D results with the maximum relative deviation in mass flow rate
around 0.85%.
4.2. The 1-2 Sub-Model
In addition to the thermo-hydraulic results comparing with
INDITEP data (Fig. 10), the 2 Sub-Model is also validated concern-
ing the circumferential distribution of temperature difference
between water/steam and the receiver tube. Actually, the practical
distribution shapes double-peak-like, labeled as practical distribu-
tion in Fig. 6. However, in calculations, simplified distributions are
commonly used. Two simplified treatments of circumferential con-

6.30 10.25

6.25
10.20
Pressure /MPa
Pressure /MPa

6.20
10.15
6.15
10.10
6.10

10.05
6.05

6.00 10.00
375
325

350

300
Temperature /
Temperature /

325

300
275

275
Lobon et al. [22] Lobon et al. [22]
250 1 Sub-Model
1 Sub-Model
250

225 225
0 100 200 300 400 500 0 100 200 300 400 500

L /m L /m

(a) (b)
Fig. 9. Validation of the 1 Sub-Model by data from DISS project [22]. (a) case 4; (b) case 5.
 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187 183

centrated solar radiation flux are considered. The more simplified

treatment of the two is to use rectangular distribution (see
Fig. 6), as was done by Eck et al. [16]. Eck et al.’s result and the pre-
sent work are compared in Fig. 11a and b, respectively, which show
a good agreement. The high temperature region clearly indicates
the part of HCE facing the reflecting mirror. The three stages of
fluid state, i.e. preheating, evaporation and superheating stages,
are also clearly seen. The maximum circumferential temperature
difference could reach up to 22–25 K. The less simplified treatment
of the two is the Gaussian distribution which is more reasonable
for thermal and thermo-elastic analysis (see Fig. 6). The result is
also shown in Fig. 11c. It is seen that the shape of profile stays con-
sistent but smoother with higher peak. The maximum circumfer-
ential temperature difference is about 23.6 K.

4.3. The 0-1-2-3 Model

According to the 2D temperature distribution (Fig. 11), the

superheating stage experiences the most thermal load. It is
believed that the most severe deformation befalls this stage. A
receiver tube close to injection port has been given a 3D results
under designed conditions. It is shown in Fig. 12a that the temper-
ature distribution of the tube and the middle section temperature
profile. Under this condition, htilt is about 6°. For the receiver tube,
the maximum temperature presents at the outer face of the end

Fig. 11. Validation of the 2 Sub-Model regarding the temperature difference

between the medium line of tube and the fluid. (a) Eck et al.’s model [16]; (b) the 2
Sub-Model with rectangular-flux distribution; (c) the 2 Sub-Model with Gaussian-
flux distribution.

Fig. 10. Validation of the 1 and 2 Sub-Models by data from INDITEP project [24].
and coincides with the tilt angle direction. The maximum temper-
ature difference at middle section is 35.45 °C. The deformation of
the tube results from the stress induced by the thermal load. To
interpret the problem, the stress distribution with full displace-
ment constrains is shown in Fig. 12b and the stress distribution
without constrains is shown in Fig. 12c. Under the constraint
184 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187
conditions, the inner and outer faces located in rim regions experi-
ence more stress and the maximum appearing at the inner face
close to the tilt direction which is 40.2 MPa, while the middle of
the two faces has lower stress. Compared with the rim regions,
the other regions bear more moderate stress. After removing the
constrains, the receiver tube bends and the stress has been largely
released. The released stress distribution seems to be more homo-
geneous and the maximum reduces to 31.2 MPa. The stress differ-
ence between the maximum and the minimum in middle section
has reduced from 39.9 MPa to 28.0 MPa. The maximum radial dis-
placement of the receiver tube presents towards the tilt angle and
the result is 1.52 cm. The corresponding glass tube displacement is
0.15 cm which is about one tenth of that of the receiver tube.
The output work from the plant is obtained directly from the 0
Sub-Model and the overall energy and exergy conversion informa-
tion is also obtained coupling with the 1 Sub-Model. Tables 6 and 7
display the energy and exergy transfer information at rated condi-
tions, respectively. Energy and exergy received by solar field are
263.3 MW and 245.4 MW, respectively. The conversion efficiencies
are 63.6% and 32.16% in solar field. For the power block, the net
electric power is 49.73 MWe, the efficiencies are 29.7% for energy
Fig. 12. Thermo-elastic analysis of a 4-m-long receiver tube. (a) Temperature distribution (unit: °C); (b) Stress distribution with fully displacement constraints (unit: Pa); (c)
Stress distribution after removing constraints (unit: Pa).
and 63.03% for exergy. The overall energy and exergy efficiencies
are 18.89% and 20.26%, respectively.
4.4. Computational performance
Normally, a complete 3D model definitely provides the most
comprehensive and accurate details for SF. However, the overall
performance demands a simplified treatment of whole system
including both SF and PB. It’s always at the cost of a great deal of
computing resource and time that makes unmatched less contribu-
tion to the industry applications. On the contrary, the proposed 0-
1-2-3 Model is much smarter and more excellent in addressing
these issues. The whole system is focused on while the necessary
details are not left behind in the meantime. To quantitatively eval-
uate the computational performance, a full 3D simulation of the
receiver tube is carried out in ANSYS Fluent. Analogue to Roldán’s
model [49], the LS-3 parabolic trough collector is employed. The
case with inlet pressure 6 MPa, inlet steam temperature 643 K,
steam flow rate 0.56 kg s1 and DNI 627 Wm2 has been employed
for comparison. The simulation settings, the boundary conditions
and the material properties all remain the same only except that
 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187 185

Table 6
Energy conversion at the rated condition.

Subsystem Received/MW Delivered/MW Loss/MW Energy efficiency/%

SF 263.3 167.46 95.84 63.6
PB 167.46 49.73 117.73 29.7
Overall 263.3 49.73 213.57 18.89

Table 7
Exergy conversion at the rated condition.

Subsystem Received/MW Delivered/MW Loss/MW Exergy efficiency/%

SF 245.4 78.9 166.5 32.16
PB 78.9 49.73 29.17 63.03
Overall 245.4 49.73 195.67 20.26

40

35

30

Time consuming /min

25

20

15

10
the 1-2 Sub-Model
3D model
5

0
4 8 12 16 20 24
Calculation length /m

Fig. 14. Contrast of computational time.

Fig. 13. Comparison of configuration of grids. (a) 3D model (FVM); (b) the 1-2 Sub-
Model.
the structured grids (see Fig. 13a) have been used instead of
unstructured grids in Roldán’s work. The residual deviation is fixed
at 106 for the energy equation and 103 for continuity and
momentum equations. The comparative results are listed in
Table 8. It is seen that the 3D simulation results agree well with
the experimental data. The maximum deviation in the thermal gra-
dient at the tube outlet is 3.65 K lower than the experimental data.
The 2 Sub-Model has also been triggered with the maximum devi-
ation in the tube thermal gradient of about 17%.
On the other hand, the computational time has been compared
between the full 3D simulation and the 1-2 Sub-Model calculation
(see Fig. 13b). The identical problem of calculating the temperature
distribution of receiver tube with a certain length is focused on.
The 1-2 Sub-Model is implemented in MATLAB R2015a and the full
3D simulation is carried out under ANSYS Fluent 16.2. The compu-
tational configurations are Intel(R) Core(TM) i5-4590 CPU, 8.00 GB
RAM and Microsoft Windows operating system (64-bit). The com-
putational time of the full 3D simulation shows exponential
growth while that of the 1-2 Sub-Model seems much less and basi-
cally keeps linear growth (see Fig. 14). The former is almost 15
times longer than that of the latter for a typical 4-m HCE and 23
times longer for a 24-m loop section. Note that the full 3D simula-
tion has been given the thermal boundary conditions on the recei-
ver tube, and once coupling with the glass tube and the
surroundings, the computational time surely becomes longer.
5. Perspective
The proposed fully-coupled multi-level analytical methodology
along with the 0-1-2-3 Model possesses multi-purpose analysis

Table 8
Comparison between the 1-2 Sub-Model and full 3D model.

Experiment [49] 3D FVM 1-2 Sub-Model ea (%)

Maximum temperature at the tube-outlet section/K 681.5 679.28 678.14 0.49%
Minimum temperature at the tube-outlet section/K 646.4 647.83 648.98 0.4%
Thermal gradient at the tube outlet/K 35.1 31.45 29.16 16.92%
Steam temperature increase/K 5.5 5.71 5.94 8%
a
Relative deviation of 1–2 Sub-Model results against experimental data
186 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187
capability, application flexibility and operational effectiveness. It
can be used in various ways, such as the solar-thermal-mechani
cal-electrical energy conversion analysis (0-1 Sub-Model),
thermo-hydraulic analysis (1-2 Sub-Model), thermo-elastic analy-
sis (2-3 Sub-Model), fast-forecast analysis (1-2-3 Sub-Model), per-
formance prediction (0-1-2-3 Model) and targeting insights (0-1-
2-3 Model). The 0-1-2-3 Model is not only suitable for the DSG
plant, but also for CSP plants with other types of HTFs, e.g. syn-
thetic oil and molten salt. With appropriate modifications to the
0-1-2-3 Model, the proposed fully-coupled multi-level analytical
methodology is readily applicable to other concentrating solar uti-
lization systems, such as the linear Fresnel CSP plant and industrial
steam supply system.
6. Conclusions
The present work proposes the fully-coupled multi-level analyt-
ical methodology to tackle the multi-level problems in CSP plants
and contributes to operationally auxiliary performance prediction
and component-orientated monitoring. This analytical methodol-
ogy is technically based on the proposed 0-1-2-3 Model with ver-
ifications presented. The results show good operating performance.
(1) The basic Sub-Models specialize in different roles. The 0 Sub-
Model centers on the heat-work conversion in PB. The 1 Sub-
Model works out the optical-thermal energy conversion and
thermo-hydraulic calculations in SF. The 2 Sub-Model
acquires the temperature distribution of receiver and glass
tubes of HCEs. The 3 Sub-Model completes the thermo-
elastic calculation for the tubes. The 2 and 3 Sub-Models
perform conditionally according to the preset thresholds.
All the Sub-Models work collaboratively to predict the plant
performance and implement targeted insight in global and
local scales.
(2) Compare with operational data from DISS and INDITEP pro-
jects and other published data, the thermo-hydraulic results
along the loop and the temperature field from Sub-Models
demonstrate satisfying precision. The thermal stress and
thermal deformation have also been investigated with the
3 Sub-Model. The plant performance is 18.89% for energy
efficiency and 20.26% for exergy efficiency under the rated
conditions. An absolute 3D model has been implemented
to compare the computational performance and the final
results demonstrate the superior merits of the 0-1-2-3
Model in computational efficiency.
(3) The fully-coupled multi-level analytical methodology offers
a fresh approach to investigate and improve the CSP technol-
ogy. This analytical methodology is well-adapted to various
CSP systems and of great significance to the application of
the CSP technology.
Acknowledgements
This work was supported by the National Natural Science Foun-
dation of China (U1261112, 51406205), the Research Project of
Chinese Ministry of Education (113055A), and the National Basic
Research Program of China (973 Program) (2015CB251505).
References
[1] R. Forristall, Heat Transfer Analysis and Modeling of a Parabolic Trough Solar
Receiver Implemented in Engineering Equation Solver, NREL/TP-550-34169,
October 2003.
[2] R.V. Padilla, G. Demirkaya, D.Y. Goswami, E. Stefanakos, M.M. Rahman, Heat
transfer analysis of parabolic trough solar receiver, Appl. Energy 88 (2011)
5097–5110.
[3] H. Liang, S. You, H. Zhang, Comparison of different heat transfer models for
parabolic trough solar collectors, Appl. Energy 148 (2015) 105–114.
[4] J. Guo, X. Huai, Multi-parameter optimization design of parabolic trough solar
receiver, Appl. Therm. Eng. 98 (2016) 73–79.
[5] J. Guo, X. Huai, Performance investigation of parabolic trough solar receiver,
Appl. Therm. Eng. 95 (2016) 357–364.
[6] S.A. Kalogirou, A detailed thermal model of a parabolic trough collector
receiver, Energy 48 (2012) 298–306.
[7] F. Zaversky, R. Medina, J. García-Barberena, Object-oriented modeling for the
transient performance simulation of parabolic trough collectors using molten
salt as heat transfer fluid, Sol. Energy 95 (2013) 192–215.
[8] G. Manzolini, A. Giostri, C. Saccilotto, P. Silva, E. Macchi, Development of an
innovative code for the design of thermodynamic solar power plants Part A:
Code description and test case, Renew Energy 36 (2011) 1993–2003.
[9] G. Manzolini, A. Giostri, C. Saccilotto, P. Silva, E. Macchi, Development of an
innovative code for the design of thermodynamic solar power plants part B:
Performance assessment of commercial and innovative technologies, Renew
Energy 36 (2011) 2465–2473.
[10] Y.-L. He, J. Xiao, Z.-D. Cheng, Y.-B. Tao, A MCRT and FVM coupled simulation
method for energy conversion process in parabolic trough solar collector,
Renew Energy 36 (2011) 976–985.
[11] Z.D. Cheng, Y.L. He, F.Q. Cui, R.J. Xu, Y.B. Tao, Numerical simulation of a
parabolic trough solar collector with nonuniform solar flux conditions by
coupling FVM and MCRT method, Sol. Energy 86 (2012) 1770–1784.
[12] A.A. Hachicha, I. Rodríguez, R. Capdevila, A. Oliva, Heat transfer analysis and
numerical simulation of a parabolic trough solar collector, Appl. Energy 111
(2013) 581–592.
[13] Y. Qiu, Y.L. He, Z.D. Cheng, K. Wang, Study on optical and thermal performance
of a linear Fresnel solar reflector using molten salt as HTF with MCRT and FVM
methods, Appl. Energy 146 (2015) 162–173.
[14] Z. Wu, S. Li, G. Yuan, D. Lei, Z. Wang, Three-dimensional numerical study of
heat transfer characteristics of parabolic trough receiver, Appl. Energy 113
(2014) 902–911.
[15] P. Wang, D.Y. Liu, C. Xu, L. Zhou, L. Xia, Conjugate heat transfer modeling and
asymmetric characteristic analysis of the heat collecting element for a
parabolic trough collector, Int. J. Therm. Sci. 101 (2016) 68–84.
[16] M. Eck, W.D. Steinmann, Modelling and design of direct solar steam generating
collector fields, J. Sol. Energy Eng. ASME 127 (2005) 371–380.
[17] Y. Wang, Q. Liu, J. Lei, H. Jin, Performance analysis of a parabolic trough solar
collector with non-uniform solar flux conditions, Int. J. Heat Mass Transf. 82
(2015) 236–249.
[18] Sourav. Khanna, Vashi Sharma, Suneet Singh, Shireesh B. Kedare, Explicit
expression for temperature distribution of receiver of parabolic trough
concentrator considering bimetallic absorber tube, Appl. Therm. Eng. 103
(2016) 323–332.
[19] F. Wang, J. Tan, L. Ma, C. Wang, Effects of glass cover on heat flux distribution
for tube receiver with parabolic trough collector system, Energy Convers.
Manag. 90 (2015) 47–52.
[20] M. Wirz, A. Steinfeld, M. Roesle, Three-dimensional optical and thermal
numerical model of solar tubular receivers in parabolic trough concentrators, J.
Sol. Energy Eng. 134 (2012) 41012.
[21] C. Marugán-Cruz, O. Flores, D. Santana, M. García-Villalba, Heat transfer and
thermal stresses in a circular tube with a non-uniform heat flux, Int. J. Heat
Mass Transf. 96 (2016) 256–266.
[22] D.H. Lobón, E. Baglietto, L. Valenzuela, E. Zarza, Modeling direct steam
generation in solar collectors with multiphase CFD, Appl. Energy 113 (2014)
1338–1348.
[23] J.J. Serrano-Aguilera, L. Valenzuela, L. Parras, Thermal 3D model for direct solar
steam generation under superheated conditions, Appl. Energy 132 (2014)
370–382.
[24] E. Zarza, M.E. Rojas, L. González, J.M. Caballero, F. Rueda, INDITEP: The first
pre-commercial DSG solar power plant, Sol. Energy 80 (2006) 1270–1276.
[25] A.M. Patnode, Simulation and Performance Evaluation of Parabolic Trough
Solar Power Plants, University of Wisconsin-Madison, 2006.
[26] M.J. Montes, A. Abánades, J.M. Martínez-Val, Performance of a direct steam
generation solar thermal power plant for electricity production as a function of
the solar multiple, Sol. Energy 83 (2009) 679–689.
[27] M.J. Montes et al., Performance analysis of an Integrated Solar Combined Cycle
using Direct Steam Generation in parabolic trough collectors, Appl. Energy 88
(2011) 3228–3238.
[28] M. Biencinto, L. González, L. Valenzuela, A quasi-dynamic simulation model for
direct steam generation in parabolic troughs using TRNSYS, Appl. Energy 161
(2016) 133–142.
[29] F.A. Al-Sulaiman, Energy and sizing analyses of parabolic trough solar collector
integrated with steam and binary vapor cycles, Energy 58 (2013) 561–570.
[30] N.B. Desai, S. Bandyopadhyay, Optimization of concentrating solar thermal
power plant based on parabolic trough collector, J. Clean. Prod. 89 (2014) 262–
271.
[31] M. Eck et al., Applied research concerning the direct steam generation in
parabolic troughs, Sol. Energy 74 (2003) 341–351.
[32] T.J. Kotas, The Exergy Method of Thermal Plant Analysis, Krieger Publications,
Butterworth-Heinemann, 1995.
[33] R. Petela, Exergy of undiluted thermal radiation, Sol. Energy 74 (2003) 469–
488.
[34] John A. Duffie, William A. Beckman, Solar Engineering of Thermal Process,
fourth ed., John Wiley & Sons Inc., Hoboken, 2013.
 L. Li et al. / Applied Thermal Engineering 118 (2017) 171–187
[35] V. Dudley, G.J. Kolb, A.R. Mahoney, T.R. Mancini, C.W. Matthews, M. Sloan,
et al., Test Results: Segs LS-2 Solar Collector, Sandia National Laboratories,
1994.
[36] Adrian. Bejan, Allan D. Kraus, Heat Transfer Handbook, John Wiley & Sons Inc.,
Hoboken, New Jersey, 2003.
[37] J. Sun, Q. Liu, H. Hong, Numerical study of parabolic-trough direct steam
generation loop in recirculation mode: characteristics, performance and
general operation strategy, Energy Convers. Manage. 96 (2015) 287–302.
[38] Y. Taitel, A.E. Dukler, A model for predicting flow regime transitions in
horizontal and near horizontal gas–liquid flow, AlChE J. 22 (1976) 47–55.
[39] K.E. Gungor, R.H.S. Winterton, A general correlation for flow boiling in tubes
and annuli, Int. J. Heat Mass Transfer 29 (1986) 351–358.
[40] K.E. Gungor, R.H.S. Winterton, Simplified general correlation for saturated flow
boiling and comparisons of correlations with data, Chem. Eng. Res. Des. 65
(1987) 148–156.
[41] M.G. Cooper, Heat flow rates in saturated nucleate pool boiling: a wide ranging
examination using reduced properties, in: J.P. Hartnett, T.F. Irvine (Eds.),
Advances in Heat Transfer, vol. 16, Academic Press, New York, 1984, pp. 157–
239.
187
[42] John R. Thome, Engineering Data Book III (Chapter 10), Wolverine Tube Inc.,
Lausanne, Switzerland, 2004–2010.
[43] Z. Rouhani, E. Axelsson, Calculation of volume void fraction in the subcooled
and quality region, Int. J. Heat Mass Transfer 13 (1970) 383–393.
[44] Adrian. Bejan, Convection Heat Transfer, John Wiley & Sons Inc., Hoboken, New
Jersey, 2013.
[45] Y. Cengel, Heat and Mass Transfer: A Practical Approach, third ed., McGraw-
Hill, Boston, 2006.
[46] John R. Thome, Engineering Data Book III (Chapter 13), Wolverine Tube Inc.,
Lausanne, Switzerland, 2004–2010.
[47] G.M.L. Gladwell, Solid mechanics and its applications, in: M. Reza Eslami,
Richard B. Hetnarski, et al. (Eds.), Theroy of Elasticity and Thermal Stresses,
Springer, Dordrecht, 2013.
[48] E. Zarza, L. Valenzuela, et al., The DISS project: direct steam generation in
parabolic trough systems. Operation and maintenance experience and update
on project status, J. Sol. Energy Eng. 124 (2002) 126–133.
[49] M.I. Roldán, L. Valenzuela, E. Zarza, Thermal analysis of solar receiver pipes
with superheated steam, Appl. Energy 103 (2013) 73–84.