Anda di halaman 1dari 40

Accepted Manuscript

Three-dimensional transient numerical model for the thermal performance of the


solar receiver

Li Xu, Wes Stein, Jin-Soo Kim, Zhifeng Wang

PII: S0960-1481(17)31259-4

DOI: 10.1016/j.renene.2017.12.055

Reference: RENE 9552

To appear in: Renewable Energy

Received Date: 25 September 2017

Revised Date: 30 November 2017

Accepted Date: 15 December 2017

Please cite this article as: Li Xu, Wes Stein, Jin-Soo Kim, Zhifeng Wang, Three-dimensional
transient numerical model for the thermal performance of the solar receiver, Renewable Energy
(2017), doi: 10.1016/j.renene.2017.12.055

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT

1 Three-dimensional transient numerical model for the


2 thermal performance of the solar receiver

3 Li Xu1, 2, *, Wes Stein2, Jin-Soo Kim2, Zhifeng Wang1

4 1 Key Laboratory of Solar Thermal Energy and Photovoltaic System of Chinese Academy of

5 Sciences, Beijing Engineering Research Center of Solar Thermal Power, Institute of Electrical

6 Engineering, Chinese Academy of Sciences, Beijing 100190, China

7 2 CSIRO Energy Technology, P.O. Box 330, Newcastle, NSW 2300, Australia

8 Abstract:
9 For solar thermal power plants, no steady-state operation occurs in view of
10 inherently transient natures of their initial and boundary conditions. So this study
11 proposes a mathematical model to perform the analysis on the transient behaviors of
12 the external solar receiver in the tower power technology. This 3D transient model was
13 established by dividing the receiver tube into discrete control volumes and then
14 applying the conservation of thermal energy to every single differential control volume.
15 In addition, this model was validated by simulating the HTF temperature distributions
16 and then comparing them with the reference results. By calculating the time-dependent
17 and non-uniform temperature fields of the receiver tube, this paper focuses attention on
18 the evolution of transient processes in several common scenarios involving the mass
19 flowrate variation, the start-up process and the occurrence of the heavy clouds above
20 the heliostat field. Particularly, the analysis of the transient thermal performance
21 highlights some noteworthy characteristics including serious problems such as the
22 corrosion, the thermal stress and the fatigue in the typical transitions, which might
23 require the control system to correspondingly adjust in time.

24 Keywords: External solar receiver; Tower technology; Molten salt; Thermal

* Corresponding author. Tel: +86 10 82547268. E-mail address: xuli_neu@126.com


1 / 38
ACCEPTED MANUSCRIPT

1 performance; 3D-transient model

3 1. Introduction
4 Since 2009, the concentrated solar power (CSP) has grown significantly and the
5 global deployment of CSP has increased to 4,500 MW. Especially, the flexibility
6 offered by CSP with the thermal energy storage is a key differentiator from variable
7 renewables such as PV and wind[1]. Also, CSP offers an integrated solution to the
8 coming decade’s global problems, i.e., climate change and associated shortage of
9 energy, water and food[2]. Solar power tower technology or the central receiver system
10 (CRS) is the current and future trend for CSP installation due to its potential to achieve
11 very high temperatures which leads to enhanced efficiency of converting heat into
12 electricity[3].
13 Additionally, molten salts are used as heat transfer/storage fluids in CSP due to
14 their main characteristics: stability at high temperatures, low vapor pressure, liquid state
15 in a large range of temperatures, ability to dissolve many inorganic and organic
16 compounds, viscosity generally low and high heat capacity per unit volume[4]. And
17 these advantages represent nowadays the most cost-effective technology for the
18 electricity generation for stand-alone solar power plants[5]. Therefore, the combination
19 of CRS and molten salts enables solar collection to be decoupled from the electricity
20 generation better than water/steam systems, yielding high capacity factors with solar-
21 only or low hybridization ratios[6].
22 There has been considerable work done on the thermal performance of the solar
23 central receiver using the molten salt as its heat transfer fluid (HTF). Zhang et al.[7]
24 developed a molten salt cavity receiver with Dymola to give a coupled calculation of
25 the temperature, velocity and pressure in the molten salt in order to be used in system
26 simulations of entire power plants; Rodríguez-Sánchez et al.[8] made a thermal analysis
27 with consideration of the circumferential and axial variations of the tube wall
28 temperature, which may be beneficial to find an optimal receiver configuration; Zhang
29 et al.[9] applied the transfer function method to a dynamic test method in order to
2 / 38
ACCEPTED MANUSCRIPT

1 estimate the thermal performance of the molten salt cavity receiver; Li et al.[10]
2 developed a global steady-state thermal model of a 100 kWt molten salt cavity receiver
3 for the design as part of the key project of the Ministry of Science and Technology of
4 People’s Republic of China; Pacheco[11] described two methods to test and evaluate
5 the receiver efficiency as a function of operating temperature and wind speed; Flores et
6 al.[12] focused on determining the time-averaged temperature profile of the salt, the
7 time-averaged temperature distribution on the walls of the tube, and the thermal stresses
8 in different configurations, and predicted a non-uniform temperature distribution with
9 strong gradients caused by the circumferential heat flux on the surface of the receiver
10 tube for the study of high thermal stresses on the pipe walls and the fluid decomposition;
11 Tehrani and Taylor[13] developed a thermal model for design/off-design performance
12 analyses of molten salt cavity receivers which might be beneficial for control strategies
13 to maximize the receiver thermal output and to analyze a CST plant without performing
14 detailed simulations.
15 Because of very expensive or even unavailable measurements of solar receivers,
16 numerous studies were carried out with the computational fluid dynamics (CFD) tool
17 to reveal the receiver design, performance improvement, extreme operating cases and
18 so on: For the external receivers, the distributed-flux models resulted in radiative heat
19 losses that were ~14% higher than the uniform-flux models [14]; Large filling velocity
20 and large temperature difference between the filling temperature and the initial tube
21 temperature should be avoided when operating cold filling[16]; A three-dimensional
22 model of a single tube was established to simulate the wall temperature of solar receiver
23 and then calculate thermal stress with temperature results by solving the thermal stress
24 equations[19]; In order to evaluate the heat balance, 3-D thermal analysis for a small
25 scale cavity receiver was employed using Fluent 15 and the amount of heat losses for
26 the three shapes was determined[20].
27 Usually, the governing equations in CFD models including the continuity,
28 momentum equation, kinetic energy, energy dissipation equation are built on the mesh,
29 which lets CFD models demand substantial computational efforts to obtain the
30 temperature distributions of the receiver and the fluid. Consequently, it is very hard to
3 / 38
ACCEPTED MANUSCRIPT

1 use CFD tools to catch up the transient heat processes of the receiver under operating
2 circumstances of varying complexity in view of their time consuming. For solving this
3 problem, this study proposed a three-dimensional transient model to simulate the
4 dynamic thermal performance of an external molten salt solar receiver. Taking into
5 account the trade-off between the precise and economical numerical requirements, the
6 governing differential equations of the transient model were established with
7 consideration of the conductive, convective and radiative heat transfer processes and
8 then were solved with the numerical integration method. Also, the temperature-
9 dependent thermophysical properties of the air, receiver tube and HTF were considered
10 during the computation process of the transient model.
11 By calculating the time-dependent and non-uniform temperature field at the axial,
12 circumferential and radial directions, the transient natures of the solar receiver between
13 different common scenarios were analyzed. Furthermore, in order to determine the
14 stability and durability of the whole system, this study focused on several significant
15 processes in operation including the mass flowrate variation, the start-up process, the
16 occurrence of the heavy clouds above the heliostat field and so on. Particularly, based
17 on the evolution of the maximum circumferential temperature of the inner surface and
18 the maximum radial temperature difference of the receiver tube, some serious problems
19 such as the corrosion, the thermal stress and the fatigue can be predicted what should
20 be avoided in the real operation. So this study will be potentially helpful in identifying
21 the design parameters, the optimum control and the operating strategy for the external
22 receiver solar power plants.
23

24 2. Mathematical Model

25 2.1 Model assumptions and general balance equation

26 The central receiver mounted at the top of a tower intercepts and absorbs the
27 concentrated solar flux from the heliostat field and then delivers this energy to the HTF.
28 Commonly, there are two flow loops in an external solar receiver and both of them start
4 / 38
ACCEPTED MANUSCRIPT

1 on the North from the inlet vessel and then end on the South to the outlet vessel[21]. A
2 flow loop includes multiple panels composed of several parallel receiver tubes and
3 these tubes are welded to inlet and outlet headers at either end[22]. This study, for
4 simplicity, assumes that no difference exists between any two vertical tubes in a panel.
5 Hence, we selected a vertical tube from each panel in a flow loop and connected them
6 in a series to make up an entire receiver tube from the inlet to outlet vessels of the solar
7 receiver. In the three-dimension model, the incident solar flux on the receiver tube is
8 supposed to vary with both axial and circumferential positions of the receiver tube.
9 Moreover, the heat conduction of the receiver tube transfers in the axial, circumferential
10 and radial directions whereas the temperature of the HTF is simplified to only vary in
11 the axial direction. Thus, the nodes and control volumes of the three-dimension
12 discretization mesh are necessary to be established over the entire receiver tube in its
13 axial, circumferential and radial directions, as shown in Fig. 1.
14

T b,NR,j,k

T b,NR,j-1,k T b,i,j,k T b,NR,j+1,k

T b,i,j-1,k T b,i,j+1,k

T b,1,j-1,k T b,1,j,k
T b,1,j+1,k (1)
15
T b,i,j,1 T b,i,j,k T b,i,j,NL

T HTF,k

16 (2)
17 Fig. 1 Nodes and control volumes in cylindrical coordinates for 3-D transient model:
18 (1) cross-section of receiver tube and (2)entire receiver tube from the inlet to outlet.
19

20 The Fig. 1 (1) shows positons of partial nodes in both circumferential and radial
21 directions on the cross section of the receiver tube. And the small shaded sectional area

5 / 38
ACCEPTED MANUSCRIPT

1 illustrates the surface of an internal control volume at the temperature 𝑇b,i,j,k and the

2 node (b, i, j, k) is located at its center. Additionally, the Fig. 1 (2) shows positons of
3 several nodes in the axial direction of the entire receiver tube. Moreover, in this study,
4 the mesh considers a three-dimensional system with coordinates r, θ and L in the
5 cylindrical coordinate system. And this mesh lets the radius of receiver tube r be
6 determined by the index i from 1 to NR, the circumferential angle θ by the index j from
7 1 to NC and the axial position L by index k from 1 to NL. So the total number of nodes
8 or control volumes NT is equal to the product of NR, NC and NL. Furthermore, the
9 index i starts from the inner surface of the receiver tube in the radial direction (for
10 instance, 1 and NR are at the inner and outer surfaces of the receiver tube, respectively);
11 the index j starts from the surface normal to the incident direction of the concentrated
12 sunlight and then increases in a clockwise direction; the index k follows the HTF flow
13 direction (for instance, 1 and NL are at the inlet and outlet of the receiver tube,
14 respectively).
15 Thus, the calculation of the surface areas in the cylindrical geometry is complicated
16 in view of their dependence on the radius r where the node is located. Especially, the
17 sectional areas of control volumes which are normal to the axial direction of the receiver
18 tube have to be calculated by Eqs.(1), (2) and (3) according to their position in the mesh.
19 For the internal control volume, the sectional area is
2 2
20 𝐴le = [(𝑟 + ∆𝑟/2) ‒ (𝑟 ‒ ∆𝑟/2) ]∆𝜃/2 = 𝑟∆𝑟∆𝜃. (1)

21 For the control volume i=1 where the inner surface of the receiver tube
22 approximates to the HTF, the sectional area is
2 2 4𝑟 + ∆𝑟
23 𝐴le = [(𝑟 + ∆𝑟/2) ‒ 𝑟 ]∆𝜃/2 = 8
∆𝑟∆𝜃. (2)

24 For the control volume i=NR where the outer surface of the receiver tube
25 approximates to the air, the sectional area is
2 2 4𝑟 ‒ ∆𝑟
26 𝐴le = [𝑟 ‒ (𝑟 ‒ ∆𝑟/2) ]∆𝜃/2 = 8
∆𝑟∆𝜃. (3)

27 Consequently, the volume of the control volume can be determined by

6 / 38
ACCEPTED MANUSCRIPT

1 {
𝐴le∆𝐿
𝑉b = 𝐴 ∆𝐿/2
le
for node 𝑘 from 2 to NL ‒ 1
for node 𝑘 = 1 or NL . (4)

2 For developing a set of governing equations, this study assumes that every single
3 control volume is treated as a lumped capacitance and has the homogeneously initial
4 and boundary conditions as well as the homogeneously thermophysical properties.
5 Then, with this information, each differential control volume is analyzed using the
6 energy balance relationship. In order to describe the evolution of receiver temperatures,
7 the heat transferred into or out of the control volume by the conduction, convection or
8 radiation needs to be taken into account with the energy storage term. Hence, for any
9 node from 1 to NT, the general governing differential equation can be determined from
10 Eq.(5).
𝑑𝑇b
11 𝑞abs + 𝑞cond = 𝑞lc + 𝑞lr + 𝑞HTF + 𝜌b𝑐b𝑉b 𝑑𝜏 (5)

12 where 𝑞abs is the rate of the heat transfer absorbed by the receiver tube from the incident

13 concentrated solar flux due to the inability of the receiver tube to have a perfectly

14 absorbing, 𝑞cond is the rate of the conductive heat transfer through the receiver tube at

15 the axial, circumferential and radial directions, 𝑞lc is the rate of the convective heat loss

16 from the receiver tube to the ambient, 𝑞lr is the rate of the radiative heat loss from the

17 receiver tube to the ambient, 𝑞HTF is the rate of the heat transfer from the receiver tube
𝑑𝑇b
18 to the HTF, and 𝜌b𝑐b𝑉b 𝑑𝜏 is the rate of the energy storage inside the control volume.

19 Note that this general governing differential equation must be taken with great care
20 for specified control volumes. While the energy balance equation for the internal
21 control volume only has the conductive heat transfer and energy storage terms, the
22 equations for the boundary control volumes are different with their positions in the
23 mesh due to multidimensional problems with complex boundaries. Therefore, unique
24 equations for boundary control volumes must be developed to establish a set of
25 governing equations.

26 The term 𝑞abs is equal to the product of the concentrated solar power on the surface

7 / 38
ACCEPTED MANUSCRIPT

1 of the receiver tube and the absorptivity of the outer surface of the receiver tube 𝛼b and

2 it is assumed to vary with the cosine of the angle in the outer surface normal to the
3 incident concentrated sunlight in this study, as expressed in Eq.(6).

{
0 for node 𝑖 from 1 to NR ‒ 1
𝛼b𝑞solcos𝜃 for node 𝑖 = NR & 0 ≪ 𝜃 ≪ 𝜋/2
4 𝑞abs = 0 for node 𝑖 = NR & 𝜋/2 < 𝜃 < 3𝜋/2 (6)
𝛼b𝑞solcos𝜃 for node 𝑖 = NR & 3𝜋/2 ≪ 𝜃 < 2𝜋

5 The convective heat loss and radiative heat loss terms only exist for boundary
6 control volumes where the outer surface of the receiver tube approximates to the air,
7 expressed as:

8 𝑞lc = 0 for node 𝑖 from 1 to NR ‒ 1 (7)

9 and

10 𝑞lr = 0 for node 𝑖 from 1 to NR ‒ 1. (8)

11 The heat from the receiver tube to the HTF only happens in boundary control
12 volumes where the inner surface of the receiver tube approximates to the HTF. Hence,
13 for Eq.(5),

14 𝑞HTF = 0 for node 𝑖 from 2 to NR. (9)

15 In Eq.(5), the rate at which energy is stored in the control volume includes the

16 differentiation term d𝑇b/d𝜏 with respect to the tube temperature. Thus, in order to

17 obtain a numerical solution to a 3-D transient model, governing differential equations


18 are carried out using the numerical integration technique. Additionally, in the following

19 subsections, the rates of the heat transfer 𝑞cond, 𝑞HTF, 𝑞lc and 𝑞lr will be described in

20 details.

21 2.2 Heat conduction through the receiver tube

22 For the control volume of the receiver tube in the 3-D transient model, heat
23 conduction components exist in the radial, circumferential, and axial directions,
24 respectively. Furthermore, for each direction, heat conduction components of an

8 / 38
ACCEPTED MANUSCRIPT

1 internal control volume come from six adjacent control volumes, as shown in Fig. 2.
2 Additionally, the total rate of the conductive heat transfer should be the sum of all heat
3 conduction components, as expressed in Eq.(10).
4

qcond,le+
a+
qcond,ci- q co n d ,r qcond,ci+

q co n d ,r
qcond,le-

a-

5
6 Fig. 2 Heat conduction analysis on the control volume for 3-D cylindrical geometry.

7 𝑞cond = 𝑞cond,ra + 𝑞cond,ci + 𝑞cond,le (10)

8 where 𝑞cond,ra is the rate at which heat conducted across the surface normal to the radial

9 direction, 𝑞cond,ci is the rate at which heat conducted across the surface normal to the

10 circumferential direction and 𝑞cond,le is the rate at which heat conducted across the

11 surface normal to the axial direction.


12 For heat conduction components, the internal and boundary control volumes
13 should be treated separately according to Eqs.(11), (12) and (13). For the control
14 volume i=1 or NR, the heat conduction at the radial direction is transferred with only
15 one adjacent control volume at one of its edge surfaces at the radial direction and the
16 other edge surface is assumed as the adiabatic surface. And it is similar to the control
17 volume k=1 or NL for the heat conduction at the axial direction. For the control volume
18 j=1 or NC, however, the heat conduction at the circumferential direction is transferred
19 with both adjacent control volumes, and it should be noted with the correct sequence of
20 control volumes in the mesh.

9 / 38
ACCEPTED MANUSCRIPT

{
𝑞𝑐𝑜𝑛𝑑,𝑟𝑎 ‒ + 𝑞𝑐𝑜𝑛𝑑,𝑟𝑎 + for node 𝑖 from 2 to NR ‒ 1
1 𝑞cond,ra = 𝑞𝑐𝑜𝑛𝑑,𝑟𝑎 + for node 𝑖 = 1 (11)
𝑞𝑐𝑜𝑛𝑑,𝑟𝑎 ‒ for node 𝑖 = NR

{
𝑞𝑐𝑜𝑛𝑑,𝑐𝑖 ‒ + 𝑞𝑐𝑜𝑛𝑑,𝑐𝑖 + for node 𝑗 from 2 to NC ‒ 1
2 𝑞cond,ci = 𝑞𝑐𝑜𝑛𝑑,𝑐𝑖 ‒ ,𝑁𝐶 + 𝑞𝑐𝑜𝑛𝑑,𝑐𝑖 + for node 𝑗 = 1 (12)
𝑞𝑐𝑜𝑛𝑑,𝑐𝑖 ‒ + 𝑞𝑐𝑜𝑛𝑑,𝑐𝑖 + ,1 for node 𝑗 = NC

{
𝑞𝑐𝑜𝑛𝑑,𝑙𝑒 ‒ + 𝑞𝑐𝑜𝑛𝑑,𝑙𝑒 + for node 𝑘 from 2 to NL ‒ 1
3 𝑞cond,le = 𝑞𝑐𝑜𝑛𝑑,𝑙𝑒 + for node 𝑘 = 1 (13)
𝑞𝑐𝑜𝑛𝑑,𝑙𝑒 ‒ for node 𝑘 = NL

4 Even at one direction, two heat transfer areas for a control volume are not the same
5 due to the cylindrical geometry. For example, the heat transfer area normal to the radial
6 direction is (𝑟 + ∆𝑟/2)∆𝜃𝛥𝐿 between two adjoining nodes i and i+1, so the conduction
7 heat rate is expressed as:
𝜆i(𝑟 + 𝛥𝑟/2)𝛥𝜃𝛥𝐿(𝑇b,i + 1,j,k ‒ 𝑇b,i,j,k)
8 𝑞𝑐𝑜𝑛𝑑,𝑟𝑎 + = 𝛥𝑟
. (14)

9 where 𝜆i is the thermal conductivity of the receiver tube between nodes i and i+1 which

10 is determined by the arithmetic average temperature between 𝑇b,i,j,k and 𝑇b,i + 1,j,k.

11 However, the heat transfer area between two adjoining nodes i and i-1 is
12 (𝑟 ‒ ∆𝑟/2)∆𝜃𝛥𝐿, so the other conduction heat rate at the radial direction is expressed
13 as:
𝜆i ‒ 1(𝑟 ‒ 𝛥𝑟/2)𝛥𝜃𝛥𝐿(𝑇b,i ‒ 1,j,k ‒ 𝑇b,i,j,k)
14 𝑞𝑐𝑜𝑛𝑑,𝑟𝑎 ‒ 𝛥𝑟
. (15)

15 Similarly, for the surface normal to the circumferential or radial direction, the rate
16 at which energy is transferred by conduction is determined by Eqs.(16) to (19),
17 respectively.
𝜆𝑗𝛥𝑟𝛥𝐿(𝑇b,i,j + 1,k ‒ 𝑇b,i,j,k)
18 𝑞𝑐𝑜𝑛𝑑,𝑐𝑖 + = 𝑟𝛥𝜃
(16)
𝜆𝑗 ‒ 1𝛥𝑟𝛥𝐿(𝑇b,i,j ‒ 1,k ‒ 𝑇b,i,j,k)
19 𝑞𝑐𝑜𝑛𝑑,𝑐𝑖 ‒ = 𝑟𝛥𝜃
(17)
𝜆𝑘𝐴le(𝑇b,i,j,k + 1 ‒ 𝑇b,i,j,k)
20 𝑞𝑐𝑜𝑛𝑑,𝑙𝑒 + = 𝛥𝐿
(18)
𝜆𝑘 ‒ 1𝐴le(𝑇b,i,j,k ‒ 1 ‒ 𝑇b,i,j,k)
21 𝑞𝑐𝑜𝑛𝑑,𝑙𝑒 ‒ = 𝛥𝐿
(19)

10 / 38
ACCEPTED MANUSCRIPT

1 2.3 Heat transferred to the HTF

2 For simplicity, this study treats the internal HTF flow as a one dimensional
3 problem while the heat exchange between the receiver tube and the HTF encompasses
4 the details of the temperature distributions of the receiver tube at the circumferential
5 and axial directions. Hence, for any control volume of the inner surface of the receiver
6 tube, the rate at which energy is transferred by the HTF convection is expressed in
7 Eq.(20).

8 𝑞HTF,1,j,k = ℎHTF,k𝐴in,1,j,k(𝑇b,1,j,k ‒ 𝑇HTF,k) (20)

9 where 𝐴in,1,j,k is the heat transfer area between the HTF and the control volume at the

10 inner surface of the receiver tube.

11 For most of the practical problems, the convective heat transfer coefficient ℎHTF,k

12 can be calculated according to the Nusselt number correlation as a function of the


13 Reynolds number and the Prandtl number. In this study, we use the classic Gnielinski
14 turbulent flow correlation to approximately fit the flow situation under consideration.

15 Thus, in the range of its validity (0.5≤PrHTF≤2000 and 3000≤Red≤5E6), the average

16 Nusselt number of the liquid phase can be determined by [23]:


(𝑓/8)(Red ‒ 1000)PrHTF
17 Nud = . (21)
1 + 12.7(𝑓/8) (PrHTF
0.5 2/3
‒ 1)

18 where the parameter 𝑓 is the Moody or Darcy friction factor, which can be estimated
19 from the Petukhov correlation[24], expressed in Eq.(22), for the fully developed single-
20 phase flow under turbulent conditions in a smooth passage, both of the Nusselt number

21 Nud and Reynolds number Red are based on the inner diameter of the receiver tube, and

22 the Prandtl number PrHTF is calculated with Eq.(23).

‒2
23 𝑓 = (0.79ln Red ‒ 1.64) (22)
𝜇HTF𝑐HTF
24 PrHTF = 𝜆HTF
(23)

25 where 𝜇HTF is the viscosity, 𝑐HTF is the specific heat capacity at the constant pressure
11 / 38
ACCEPTED MANUSCRIPT

1 and 𝜆HTF is the thermal conductivity of the HTF.

2 Consequently, the heat transfer coefficient for the HTF under the turbulent
3 condition can be obtained from
𝜆HTFNu
d
4 ℎHTF = 𝑑
. (24)

5 Actually, with the exception of the HTF temperature at the inlet of the receiver
6 tube provided by the operation, the HTF temperature distribution is unknown. So it
7 necessarily provides another independent relation between the HTF temperature and
8 the heat transfer rate in order to complete the set of the governing differential equations.
9 Then, for the HTF node k between 2 and NL, the energy balance is established as the
10 Eq.(25) in view of the fact that the heat carried into the next control volume by the HTF
11 flow is equal to the heat gained by the HTF from the inner surface of the receiver tube.
𝑁𝐶
12 𝑐HTF,k𝑚(𝑇HTF,k + 1 ‒ 𝑇HTF,k) = ∑𝑗 = 1𝑞HTF,1,j,k (25)

13 2.4 Convective heat loss

14 Both the natural convection caused by the density differences in the air
15 surrounding the heated solar receiver and the forced convection caused by the wind

16 flow contribute to the convective heat loss 𝑞lc from the external receiver to the ambient,

17 as shown in Eq. (26).

18 𝑞lc = ℎlc𝐴ou(𝑇b ‒ 𝑇a) (26)

19 where 𝐴ou is the heat transfer area between the air and the control volume at the outer

20 surface of the receiver tube, 𝑇a is the ambient air temperature, and the convection heat

21 loss coefficient ℎlc is a combination of the forced and natural convection effects, which

22 is estimated with the correlations in the following for the external solar receiver in terms
23 of dimensionless parameters such as the Nusselt number Nu, the Reynolds number Re
24 and the Grashof number Gr.

12 / 38
ACCEPTED MANUSCRIPT

1 For the natural convection heat loss, the heated air surrounding the receiver is
2 assumed to be turbulent (the Grashof number is above 1E10) over most of the height of
3 the surface in view of the large size of the receiver and the high operating temperatures.
4 Thus, the empirical correlation for the Nusselt number is expressed in Eq. (27) with
5 consideration of the curvature of the cylinder effect [25], which fits the experimental
6 data within ±6%[26].
1/3 0.14
7 NuH = 0.049πGr
H
( 𝑇a/𝑇b) (27)

8 where the Grashoff number GrH given by Eq. (28) is based on the height of the solar

9 receiver in the direction of the gravity H.


3
𝑔𝛽(𝑇b ‒ 𝑇a)𝐻
10 GrH = 2 (28)
𝜐a

11 where the volumetric thermal expansion coefficient β is close to 1/𝑇a for the air and the

12 kinematic viscosity of the air 𝜐a is evaluated at the ambient temperature 𝑇a.

13 For the forced convection heat loss, the Nusselt number has a relationship with the
14 effective sand grain roughness which is the ratio of the receiver tube radius D/2 by the
15 solar receiver diameter DR. And it is calculated using the correlations in Table 1 as well
16 as the Eq. (29). Additionally, all thermophysical properties of dimensionless numbers
17 for the forced convection calculation should be evaluated at the arithmetic average of
18 the tube temperature and the ambient temperature.
19 Table 1 Relationship between the roughness and the Nusselt number[27]

roughness 𝑚
NuDR = cRe Relow<ReDR<4E6
DR

D/2/DR c m Relow

0 9.02E-4 1.01 1.5E6

7.5E-4 2.57E-3 0.98 1E6

3E-3 1.35E-2 0.89 3E5

9E-3 4.55E-2 0.81 3E5


0.63
20 NuDR = 0.18Re ( when 1E4<ReDR≤Relow) (29)
DR

13 / 38
ACCEPTED MANUSCRIPT

1 where the characteristic length for both NuDR and ReDR is the solar receiver diameter

2 DR, and Relow is the low limit of the Reynolds number range for each roughness

3 parameter.
4 Based on the two Nusselt numbers above, the natural and forced convection heat
5 loss coefficients are determined by Eqs. (31) and (30), respectively.
𝜆aNu
H
6 ℎnc = 𝐻
(30)
𝜆a,bNu
DR
7 ℎfc = 𝐷𝑅
(31)

8 where the thermal conductivity for the natural convection 𝜆a is evaluated just at the

9 ambient temperature whereas the thermal conductivity for the forced convection 𝜆a,b is

10 evaluated at the arithmetic average of the tube temperature and the ambient temperature.
11 Thus, for the external receivers, the combined convective heat transfer coefficient
12 is determined by[28]
3.2 1/3.2
13 ( 3.2
ℎlc = ℎ nc + ℎ fc ) . (32)

14 2.5 Radiative heat loss

15 This study uses a simplified radiation model to estimate the thermal radiation from
16 the receiver tube to the surroundings. On the assumption that the adjacent tubes in a
17 panel have the same temperature distribution. Therefore, the radiative heat transfer
18 between the adjacent tubes in a panel can be negligible. Thus, the rate of the radiative
19 heat loss is calculated by the Eq. (33).

20 𝑞lr = 𝜀𝜎𝐹𝑏 ‒ 𝑎𝐴 𝑇b ‒ 𝑇a
lr
( 4
)
4
(33)

21 where 𝜀 is the emissivity of the outer surface of the receiver tube, 𝜎 is the Stefan-

22 Boltzmann constant, 𝐹𝑏 ‒ 𝑎 is the view factor from the convex outer surface to the

23 surroundings and 𝐴lr is the radiative heat transfer area between the surroundings and

24 the control volume at the outer surface of the receiver tube.


25

14 / 38
ACCEPTED MANUSCRIPT

1 3. Results and discussion

2 3.1 Validation

3 To verify the reliability of the 3-D transient model for simulating the thermal
4 performance of the external solar receiver, this study uses almost the same geometry,
5 material and boundary conditions specified in the published reference [29] as follows:
6 A cylindrical receiver has the outer diameter DR = 8.5 m and the vertical length H
7 = 10.5 m. The receiver tube has an outer diameter D = 45 mm with a wall thickness th
8 = 1.5 mm. There are nine panels in a flow loop. And one vertical tube is chosen from
9 each panel and then these nine vertical tubes are connected in a series to make up an
10 entire receiver tube described in the previous section (2.1 Model assumptions and
11 general balance equation). Thus, the total length of this entire receiver tube is 94.5 m.
12 In addition, the absorptivity and emissivity of the receiver tube with a selected coating
13 are 0.93 (effective over the life time)[21] and 0.87, respectively. Moreover, the typical
14 binary salt mixture (60% NaNO3 and 40% KNO3 by weight) is used as the HTF, and its
15 thermophysical properties were given in the reference [30]. The external surface of the
16 solar receiver is exposed to the non-uniform solar flux as illustrated in Fig. 3 and to the

17 air at 𝑡a = 35 °C. This concentrated solar flux is reflected from the heliostat field and

18 the direct normal irradiance (DNI) is 910 W/m2. In addition, for the solar flux
19 distribution on a single vertical tube in a panel, the lower values occur in two end
20 positions of this vertical tube at the axial direction. Consequently, the solar flux at the
21 axial direction on the entire receiver tube in Fig. 3 is obviously lower at 0 m, 10.5 m,
22 21 m, 31.5 m, 42 m, 52.5 m, 63 m, 73.5 m, 84 m, and 94.5 m because this entire receiver
23 tube consists of nine vertical tubes in a series from nine different panels as described
24 above. Furthermore, the molten salt at 290°C enters the inlet of the receiver with the
25 mass flow rate 𝑚 = 4.22 kg/s.

15 / 38
ACCEPTED MANUSCRIPT

1
2 Fig. 3. Concentrated solar flux distribution on the receiver tube.
3 Consequently, based on the information described above, the temperature
4 distribution of the receiver tube calculated by the 3-D transient model is shown in Fig.
5 4. It demonstrates that the high temperature appears on the control volumes near the
6 outlet of the receiver due to the heated HTF although the high solar flux mainly exits
7 on the receiver surface close to the inlet. In Fig. 5, the temperature distribution on the
8 cross-section of the receiver tube at the outlet shows that the maximum temperature on
9 the cross-section occurs at the position 𝜃 = 0° normal to the concentrated sun lights
10 while the control volumes without the incident solar flux in 90°≪θ≪270° are almost at
11 the same relatively low temperatures.

16 / 38
ACCEPTED MANUSCRIPT

1
2 Fig. 4. 3D temperature distribution of the receiver tube.

3
4 Fig. 5. Temperature distribution on the cross section of the receiver tube at the outlet.
5 In order to compare with the steady-state results in the reference, this study selects
6 the temperature distributions at the time when the receiver achieves the stationary
7 equilibrium after the transient process ends. As shown in Fig. 6, the HTF temperature
8 distribution along the flow direction simulated by the transient model has a deviation

17 / 38
ACCEPTED MANUSCRIPT

1 of no more than ±6 °C from the reference calculation. Although it seems that the
2 temperature differences are larger in the center than in other position, the major reasons
3 for it might be that the figures of calculation conditions as well as results in the reference
4 cannot be read precisely, and errors are caused by the transformation of the solar flux
5 from one dimensional distribution along the length of the receiver into the curve surface
6 distribution as illustrated in Fig. 3. Hence, the satisfactory agreement indicates that the
7 3-D transient model is reasonable and reliable.
6

4
Temperature (  C)

-1

-2

-3
0 10 20 30 40 50 60 70 80 90 100

8 Length position (m)

9 Fig. 6. HTF temperature differences between the model results and those of the
10 reference.

11 3.2 Transient performance analysis

12 In reality, temperature distributions immediately respond to any suddenly changed


13 condition, then start to experience a transient process and finally achieve a stationary
14 equilibrium again when the temperature of any control volume does not change from
15 the previous time step. And the 3D transient model has the ability to identify how the
16 temperature distributions evolve from the initial condition to another new equilibrium
17 after the receiver is exposed to varied conditions such as the solar flux, the HTF flowrate,
18 the wind speed and the ambient temperature. In the previous subsection, the simulation
19 in the reference was carried out without the initial value problem due to its steady-state
20 model. On the contrary, the initial condition must be prescribed for a transient model.
21 Actually, the temperature distribution of the receiver tube at the time τ = 0 s for the 3-

18 / 38
ACCEPTED MANUSCRIPT

1 D transient model in the previous subsection is a uniform distribution at 260 °C which


2 may be a start-up operation condition in the solar tower power plant.
3 As shown in Fig. 7, the maximum temperature on the cross-section of the receiver
4 tube as a function of the length position and the simulation time evolves from a uniform
5 initial temperature distribution at 260 °C. It clearly demonstrates that the time lag
6 caused by the thermal capacity of the receiver tube after a sudden incidence of the
7 concentrated sun’s rays can be simulated by the transient model. Also, it illustrates that
8 the temperature distribution along the length of the receiver tube strongly varies in
9 several time steps at the beginning. And this distribution tends to become the fluctuating
10 quantity with the peak and valley along the length of the receiver tube which eventually
11 coincides with the distribution shape of the concentrated solar flux at axial direction in
12 Fig. 4. Moreover, it takes about 35 s to experience the corresponding transient process
13 and then the receiver tube temperature achieves a stationary equilibrium.
14 In contrast to this distribution of the maximum temperature on the cross-section,
15 as shown in Fig. 8, the larger values appear in the positions close to the inlet of the
16 receiver tube for the difference between the maximum and minimum temperatures on
17 the same cross-section of the receiver tube, which has a very similar trend with the solar
18 flux distribution in Fig. 4. And Fig. 8 also shows that the temperature differences for
19 some positions initially increase but subsequently reduce until they achieve the
20 stationary equilibrium as a result of the complicated combination of the local solar flux
21 and the internal HTF flow.
22 In addition, Fig. 9 illustrates that what distribution of the maximum temperature
23 difference at radial direction of the receiver tube is at the simulation time 1 s, 3 s, 10 s
24 and 60 s, respectively after the receiver tube was exposed to the concentrated solar flux
25 as illustrated in Fig. 3. As described before, the entire receiver tube consists of nine
26 vertical tubes (the length of a single vertical tube is 10.5 m) in a series and the lower
27 values of the solar flux occur in two end positions of the vertical tube at the axial
28 direction. And because of these lower values of the solar flux at 0 m, 10.5 m, 21 m,
29 31.5 m, 42 m, 52.5 m, 63 m, 73.5 m, 84 m, and 94.5 m, the maximum radial temperature
30 differences at radial direction are plainly lower at these positions. Furthermore, in this
19 / 38
ACCEPTED MANUSCRIPT

1 study, the radial temperature difference which may reflect the thermal stress is defined
2 as the temperature difference between the outer and inner surfaces at the same axial and
3 circumferential positions of the receiver tube. Moreover, for the initially uniform
4 temperature distribution, the radial temperature difference is 0 °C at the time = 0 s,
5 which is changed rapidly with new boundary conditions. Similar to the distribution of
6 the maximum temperature difference on the cross-section, the larger radial temperature
7 differences occur in the positions close to the inlet of the receiver tube and arrive at
8 approximately 60 °C after the simulation time = 10 s, which means that the largest
9 thermal stress may occur in the first two panels of the solar receiver.
10

11

12 Fig. 7. Evolution of the maximum temperature on the cross-section of the receiver


13 tube.

20 / 38
ACCEPTED MANUSCRIPT

1
2 Fig. 8. Evolution of the maximum temperature difference on the cross-section of the
3 receiver tube.
4

60
1s
3s
10 s
50 60 s

40
Temperature (  C)

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Length position (m)
5
6 Fig. 9. Evolution of the maximum radial temperature difference.
21 / 38
ACCEPTED MANUSCRIPT

1 3.3 Flow rate variation influence

2 The variation in the flow rate of the HTF commonly occurs in the on-site operation
3 of solar thermal power plants. It is therefore necessary to explain how this variation
4 affects the transient thermal performance of the receiver. For the comparison in
5 difference cases, this study selects a design operation point as the basic case: the air

6 temperature 𝑡a = 35 °C, the DNI = 910 W/m2 with the solar flux distribution in Fig. 4,

7 and the mass flow rate 𝑚 = 4.22 kg/s which let the outlet temperature of the molten salt
8 approach 565 °C when the inlet temperature is 290 °C. Moreover, the material of the
9 receiver tube in this study is made of the Incoloy Alloy 800H for which the maximum
10 inner surface temperature is 630 °C because molten nitrate salts become corrosive at
11 this high temperature[29].
12 In Fig. 10 (1), with the mass flow rate reducing to 70 %, the maximum
13 circumferential temperature of the inner surface arises continuously from that of the
14 design operation point in the previous subsection. Especially, those temperatures of the
15 control volumes near the outlet of the receiver tube can achieve more than 630 °C at
16 the time = 3 s. As a result, in consideration of the safety of the solar receiver system,
17 this case should be avoided in the real operation which requires the control system to
18 monitor this kind of the variation and correspondingly adjust the mass flow rate in about
19 3 s after the mass flow rate drops by 30 %. Additionally, in Fig. 10 (2), the maximum
20 radial temperature difference changes immediately, but then the minor variation in it
21 occurs in the following time. Furthermore, in comparison with the distribution at the
22 time = 60 s as shown in Fig. 9, the maximum radial temperature difference goes down
23 with the reduction of 30 % in the mass flow rate.
24 In contrast to the case above, as illustrated in Fig. 11 (1), with the mass flow rate
25 increasing by 30 %, the maximum circumferential temperature of the inner surface
26 reduces continuously from that of the design operation point. But similar to the case
27 above, those temperatures of the control volumes near the outlet of the receiver tube
28 change markedly. In addition, compared with the distribution at the time = 60 s as

22 / 38
ACCEPTED MANUSCRIPT

1 shown in Fig. 9, the Fig. 11 (2) shows that the maximum radial temperature difference
2 increases immediately from that of the design operation point at the time = 1s. However,
3 some values of the control volumes near the inlet of the receiver tube reduce in the
4 following time until the system achieves a new steady state as a result of the
5 complicated combination of the initial temperature and the variation in the flow rate of
6 the molten salt.

750

700

650

600
Temperature (  C)

550

500

450

400

1s
350 3s
10 s
60 s
300
0 10 20 30 40 50 60
Length position (m)
70 80 90
(1) 100

60
1s
3s
10 s
50 60 s

40
Temperature (  C)

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Length position (m) (2)
8
9 Fig. 10. Evolution of tube temperature with 0.7𝑚: (1) maximum circumferential
10 temperature of the inner surface and (2) maximum radial temperature difference.

23 / 38
ACCEPTED MANUSCRIPT

650

600

Temperature (  C) 550

500

450

400

350 1s
3s
10 s
60 s
300
0 10 20 30 40 50 60
Length position (m)
70 80 90
(1) 100

70
1s
3s
10 s
60 60 s

50
Temperature (  C)

40

30

20

10

(2)
0 10 20 30 40 50 60 70 80 90 100
Length position (m)
2
3 Fig. 11. Evolution of tube temperature with 1.3𝑚: (1) maximum circumferential
4 temperature of the inner surface and (2) maximum radial temperature difference.
5

6 In reality, a common scenario for an operating solar thermal power plant is that,
7 for the purpose of the required HTF temperature at the receiver outlet, the HTF flow
8 rate is adjusted according to the varying level of the solar irradiance. Additionally, the
9 concentrated solar flux distribution over the receiver surface depends on the position of
10 the sun, the efficiency of the concentrating mirrors and so on. However, for

24 / 38
ACCEPTED MANUSCRIPT

1 convenience, this study uses a fairly simplistic approach to obtain the concentrated solar
2 flux distributions for different DNI levels; that is, for any position of the receiver
3 surface, the solar flux is equal to the product of the solar flux at the same position in
4 the design operation point and the proportion of the DNI to 910 W/m2.
5 Different from the specific conditions at the design operation point are the DNI =
6 600 W/m2 and the mass flow rate 𝑚 = 2.6 kg/s which let the outlet temperature of the
7 molten salt approach 565 °C. And as shown in Fig. 12 (1), similar to the transient
8 process in the previous subsection, the maximum circumferential temperature of the
9 inner surface evolves from a uniform temperature distribution at 260 °C. At the start,
10 such as the time = 3 s, the positons close to the inlet of the receiver tube turn to the
11 higher temperatures than those close to the outlet. But the control volumes close to the
12 inlet attain the new energy balance firstly and then their temperatures stop increasing.
13 While the temperatures of the control volumes close to the outlet still keep rising until
14 the whole transient process completes and become a non-uniform distribution at the
15 time = 60 s. In addition, Fig. 12 (2) illustrates that the maximum radial temperature
16 difference increases immediately from zero due to the uniform initial temperatures. And
17 just at the time = 1 s it achieves more than 70 % of its ultimately steady value when the
18 transient process ends. Furthermore, compared with the distribution at the time 60 s
19 shown in Fig. 9, the ultimate distribution of the maximum radial temperature difference
20 after the receiver experiences the transient process is lower because of the lower DNI
21 level even though the mass flow rate is reduced to make the molten salt have the same
22 outlet temperature 565 °C.
23 In another case, different from the specific conditions at the design operation point
24 are the DNI = 1200 W/m2 and the mass flow rate 𝑚 = 5.7 kg/s which let the outlet
25 temperature of the molten salt approach 565 °C. In comparison with the evolution in
26 Fig. 12 (1), the maximum circumferential temperature of the inner surface responds
27 more quickly and increases faster at a higher DNI, as shown in Fig. 13 (1). However,
28 the ultimate temperatures for both cases are very close as a result of the similar HTF
29 temperatures at the adjusted flow rates. Additionally, Fig. 13 (2) demonstrates that the
30 maximum radial temperature differences of the control volumes near the outlet of the
25 / 38
ACCEPTED MANUSCRIPT

1 receiver tube increase at the time = 1 or 3 s, then reduce at the time = 10 s, and
2 eventually increase again, which might result in thermal fatigue problem. Moreover,
3 compared with the distribution at the time = 60 s as shown in Fig. 9, the ultimate
4 distribution of the maximum radial temperature difference is higher in the view of the
5 higher DNI level even if the adjusted mass flow rate makes sure the outlet temperature
6 of the molten salt at about 565 °C So this means that the higher DNI level might cause
7 the larger thermal stress even though the larger mass flow rate maintains the same outlet
8 temperature of the molten salt.

650

600

550
Temperature (  C)

500

450

400

350

1s
300
3s
10 s
60 s
250
0 10 20 30 40 50 60
Length position (m)
70 80 90
(1) 100

40
1s
3s
35 10 s
60 s

30
Temperature (  C)

25

20

15

10

0
0 10 20 30 40 50 60
Length position (m)
70 80 90
(2)100
10

26 / 38
ACCEPTED MANUSCRIPT

1 Fig. 12. Evolution of tube temperature with DNI =600W/m2: (1) maximum
2 circumferential temperature of the inner surface and (2) maximum radial temperature
3 difference.
4

650

600

550
Temperature (  C)

500

450

400

350

1s
300
3s
10 s
60 s
250
0 10 20 30 40 50 60
Length position (m)
70 80 90
(1) 100

80
1s
3s
70 10 s
60 s

60
Temperature (  C)

50

40

30

20

10

(2)
0 10 20 30 40 50 60 70 80 90 100
Length position (m)
6
7 Fig. 13. Evolution of tube temperature with DNI =1200W/m2: (1) maximum
8 circumferential temperature of the inner surface and (2) maximum radial temperature
9 difference.

27 / 38
ACCEPTED MANUSCRIPT

1 3.4 DNI variation influence

2 Since the solar energy belongs to unstable resources, the variation in the DNI
3 always happens even on sunny days. Consequently, the transient thermal behaviors of
4 the receiver have significant effects on the system safe and operation strategy if the
5 HTF flow rate is not adjusted in time when the receiver tube is suddenly exposed to a
6 new solar flux. So, in this study, three different DNI levels (1100, 600 and 0 W/m2) are
7 proposed as the only varied conditions in order to analyze how the temperatures of the
8 receiver tube evolve from those at the design point specified above.
9 Fig. 14 (1) suggests that the maximum inner surface temperature instantly
10 increases when the receiver tube is exposed a higher DNI level than that of the design
11 operation point. And those temperatures of the control volumes near the outlet of the
12 receiver tube achieve more than 630 °C at the time = 3 s. As a result, in consideration
13 of the serious corrosion problem of the material of the receiver tube, this case should
14 be avoided in the real operation. And thus it requires the control system to monitor this
15 kind of the variation in time and correspondingly adjust the heliostat field or the mass
16 flow rate in about 3 s after the DNI arises to 1100 W/m2. In addition, as shown in Fig.
17 14 (2), the variation in the maximum radial temperature difference almost is completed
18 at the time = 1 s with the higher DNI level. Furthermore, compared with the distribution
19 at the time = 60 s shown in Fig. 9, the ultimate distribution of the maximum radial
20 temperature difference after the receiver experiences the transient process becomes
21 higher.
22 If sometimes the heavy clouds appearing above the heliostat field block the sun
23 lights, then the solar flux incident on the receiver tubes will instantly turn to zero. In
24 this case, Fig. 15 (1) shows the maximum inner surface temperatures essentially change
25 from those of the design operation point. Especially, the distribution near the outlet of
26 the receiver tube tends to be smooth quickly as the molten salt progresses through the
27 receiver tube. Finally, the maximum inner surface temperatures at the time = 60 s are
28 less than the HTF inlet temperature due to the heat loss. In addition, the maximum radial
29 temperature difference also reduces until the negative values occur. And it means that
28 / 38
ACCEPTED MANUSCRIPT

1 the inner surface temperature of the receiver tube is larger than the outer surface
2 temperature due to the relatively high HTF temperature, as shown in Fig. 15 (2).
3 Moreover, it should be noted that the control volumes near the outlet are heated at the
4 time = 10 s because of the energy carried by the HTF flow from the control volumes
5 near the inlet with the higher temperatures.
6

700

650

600
Temperature (  C)

550

500

450

400

1s
350
3s
10 s
60 s
300
0 10 20 30 40 50 60
Length position (m)
70 80 90
(1) 100

70
1s
3s
10 s
60 60 s

50
Temperature (  C)

40

30

20

10
0 10 20 30 40 50 60
Length position (m)
70 80 90
(2)100
8

29 / 38
ACCEPTED MANUSCRIPT

1 Fig. 14. Evolution of tube temperature with DNI =1100W/m2: (1) maximum
2 circumferential temperature of the inner surface and (2) maximum radial temperature
3 difference.
4

600

550

500
Temperature (  C)

450

400

350

300 1s
3s
10 s
60 s
250
0 10 20 30 40 50 60
Length position (m)
70 80 90
(1) 100

20
1s
3s
10 s
60 s
15
Temperature (  C)

10

-5

(2)
0 10 20 30 40 50 60 70 80 90 100
Length position (m)
7
8 Fig. 15. Evolution of tube temperature with DNI =0W/m2: (1) maximum
9 circumferential temperature of the inner surface and (2) maximum radial temperature
10 difference.

30 / 38
ACCEPTED MANUSCRIPT

2 4. Conclusion
3 In this study, a 3D transient mathematical model was developed to precisely
4 describe how thermal behaviors of the external solar receiver evolve with varying initial
5 and boundary conditions. The solution of this model was derived using the numerical
6 integration method with an adaptive step-size to make sure its reliability and accuracy.
7 In addition, two essential factors taken into account were the mass flow rate of the HTF
8 and the DNI level. Then, a comprehensive analysis of how they influence the transient
9 processes of solar receivers was carried out, which obtains the maximum
10 circumferential temperature of the inner surface and the maximum radial temperature
11 difference between the inner and outer surfaces.
12 Consequently, there are several important conclusions from this analysis:
13 (1). For cases with the variation of ±30 % in the mass flow rate, the inner surface
14 temperatures of the receiver tube near the outlet will change markedly; Especially, a 30
15 % reduction in the mass flow rate causes the receiver tube to be more than 630 °C at 3
16 s, which should be avoided by the adjustment in time because of the corrosion of the
17 receiver material caused by the molten salt; Furthermore, the ultimate maximum radial
18 temperature difference becomes smaller with the reduction in the mass flow rate.
19 (2). For cases with the adjustment of the mass flow rate for the variation in the
20 DNI, the maximum circumferential temperature of the inner surface responds more
21 quickly and increases faster at a higher DNI but no more than 630 °C occurs; The
22 ultimate maximum radial temperature differences after the receiver experiences the
23 transient process become larger at a higher DNI, which might cause the larger thermal
24 stress even though the mass flow rate is adjusted to make the molten salt have the same
25 outlet temperature 565 °C; Additionally, receiver temperatures near the outlet do not
26 change gradually, which might result in thermal fatigue problem.
27 (3). For cases with the variation in the DNI from the operation design point without
28 the mass flow rate adjustment, receiver temperatures near the outlet of the receiver will
29 achieve more than 630 °C at 3 s after the DNI arises from 910 W/m2 to 1100 W/m2;

31 / 38
ACCEPTED MANUSCRIPT

1 The maximum radial temperature differences near the outlet of the receiver do not
2 change gradually as a result of the complicated combination of the initial temperature
3 and the flow of the molten salt; Moreover, the maximum radial temperature difference
4 reduces successively until the negative values occur due to the relatively high HTF
5 temperature when the solar flux incident on the receiver instantly turns to zero.
6

7 Acknowledgment
8 This work was supported by the National Natural Science Foundation of China
9 (No. 51406194), the Australian Renewable Energy Agency (ARENA), and China
10 Scholarship Council.
11

12 Nomenclature
A area, m2

c specific heat capacity at constant pressure (J/kg K)

D outer diameter of the receiver tube (m)

DR outer diameter of the external receiver(m)

d inner diameter of the receiver tube (m)

f friction factor

F view factor

Gr Grashof number

g gravitational constant (9.81m/s2)

H height of the solar receiver (m)

h convection heat transfer coefficient (W/m2 K)

L length of the receiver tube (m)

𝑚 mass flow rate (kg/s)

Nu Nusselt number

32 / 38
ACCEPTED MANUSCRIPT

Pr Prandtl number

q rate of heat transfer (W)

r radius of the receiver tube (m)

Re Reynolds number

T temperature (K)

t temperature (°C)

th thickness of receiver tube (m)

u wind speed (m/s)

V volume (m3)

Greek symbols

α absorptivity

β volumetric thermal expansion coefficient (1/K)

ε emissivity

η overall energy percentage (s)

θ Circumferential angle of the receiver tube (°)

σ Stefan-Boltzmann constant (5.67E-8 W/m2 K4)

λ thermal conductivity (W/kg K)

μ viscosity (Pa s)

ρ density (kg/m3)

τ Time (s)

υ kinematic viscosity (m2/s)

Subscripts

a ambient

abs solar flux absorbed by the receiver tube

33 / 38
ACCEPTED MANUSCRIPT

an annulus zone of the receiver tube

b receiver tube

cond conductive heat transfer

ci circumference direction

fc forced convection

nc natural convection

lc convective heat loss

le length direction

lr radiative heat loss

HTF heat transfer fluid

i node index in the radial direction

in inner surface of the receiver tube

ini initial condition

j node index in the circumferential direction

k node index in the axial direction

NC total number of the nodes in the circumference

NL total number of the nodes in the length

NR total number of the nodes in the radius

ou outer surface of the receiver tube

ra radius direction

sol solar

st steady-state

tr transient

34 / 38
ACCEPTED MANUSCRIPT

1 Reference

2 [1] M. Mehos, C. Turchi, J. Jorgenson, P. Denholm, C. Ho, K. Armijo, On the Path to

3 SunShot: Advancing Concentrating Solar Power Technology, Performance, and

4 Dispatchability, National Renewable Energy Lab. (NREL), Golden, CO (United States),

5 2016, p. Medium: ED; Size: 66 p.

6 [2] O. Behar, A. Khellaf, K. Mohammedi, A review of studies on central receiver solar

7 thermal power plants, Renewable and Sustainable Energy Reviews 23 (2013) 12-39.

8 [3] K. Vignarooban, X. Xu, A. Arvay, K. Hsu, A.M. Kannan, Heat transfer fluids for

9 concentrating solar power systems – A review, Appl Energ 146 (2015) 383-396.

10 [4] V.M.B. Nunes, C.S. Queirós, M.J.V. Lourenço, F.J.V. Santos, C.A. Nieto de Castro,

11 Molten salts as engineering fluids – A review: Part I. Molten alkali nitrates, Appl Energ

12 183 (2016) 603-611.

13 [5] J.M. Lata, M. Rodríguez, M.Á. de Lara, High Flux Central Receivers of Molten

14 Salts for the New Generation of Commercial Stand-Alone Solar Power Plants, Journal

15 of Solar Energy Engineering 130(2) (2008) 021002-021002.

16 [6] J.I. Ortega, J.I. Burgaleta, F.M. Téllez, Central Receiver System Solar Power Plant

17 Using Molten Salt as Heat Transfer Fluid, Journal of Solar Energy Engineering 130(2)

18 (2008) 024501-024501.

19 [7] Q. Zhang, X. Li, Z. Wang, J. Zhang, B. El-Hefni, L. Xu, Modeling and simulation

20 of a molten salt cavity receiver with Dymola, Energy 93, Part 2 (2015) 1373-1384.

21 [8] M.R. Rodríguez-Sánchez, A. Soria-Verdugo, J.A. Almendros-Ibáñez, A. Acosta-

22 Iborra, D. Santana, Thermal design guidelines of solar power towers, Appl Therm Eng

35 / 38
ACCEPTED MANUSCRIPT

1 63(1) (2014) 428-438.

2 [9] Q. Zhang, X. Li, Z. Wang, C. Chang, H. Liu, Experimental and theoretical analysis

3 of a dynamic test method for molten salt cavity receiver, Renew Energ 50 (2013) 214-

4 221.

5 [10] X. Li, W. Kong, Z. Wang, C. Chang, F. Bai, Thermal model and thermodynamic

6 performance of molten salt cavity receiver, Renew Energ 35(5) (2010) 981-988.

7 [11] J.E. Pacheco, Final Test and Evaluation Results from the Solar Two Project, Sandia

8 National Laboratories, Albuquerque, New Mexico, USA, 2002.

9 [12] O. Flores, C. Marugán-Cruz, D. Santana, M. García-Villalba, Thermal Stresses

10 Analysis of a Circular Tube in a Central Receiver, Energy Procedia 49 (2014) 354-362.

11 [13] S.S. Mostafavi Tehrani, R.A. Taylor, Off-design simulation and performance of

12 molten salt cavity receivers in solar tower plants under realistic operational modes and

13 control strategies, Appl Energ 179 (2016) 698-715.

14 [14] J.M. Christian, C.K. Ho, CFD simulation and heat loss analysis of the solar two

15 power tower receiver, ASME 2012 6th International Conference on Energy

16 Sustainability collocated with the ASME 2012 10th International Conference on Fuel

17 Cell Science, Engineering and Technology, American Society of Mechanical Engineers,

18 2012, pp. 227-235.

19 [15] Z. Liao, X. Li, C. Xu, C. Chang, Z. Wang, Allowable flux density on a solar central

20 receiver, Renew Energ 62 (2014) 747-753.

21 [16] Z. Liao, X. Li, Z. Wang, C. Chang, C. Xu, Phase change of molten salt during the

22 cold filling of a receiver tube, Sol Energy 101 (2014) 254-264.

36 / 38
ACCEPTED MANUSCRIPT

1 [17] Z. Liao, A. Faghri, Thermal analysis of a heat pipe solar central receiver for

2 concentrated solar power tower, Appl Therm Eng 102 (2016) 952-960.

3 [18] C. Chang, C. Xu, Z.Y. Wu, X. Li, Q.Q. Zhang, Z.F. Wang, Heat Transfer

4 Enhancement and Performance of Solar Thermal Absorber Tubes with

5 Circumferentially Non-uniform Heat Flux, Energy Procedia 69 (2015) 320-327.

6 [19] B.-C. Du, Y.-L. He, Z.-J. Zheng, Z.-D. Cheng, Analysis of thermal stress and

7 fatigue fracture for the solar tower molten salt receiver, Appl Therm Eng 99 (2016)

8 741-750.

9 [20] A.M. Daabo, S. Mahmoud, R.K. Al-Dadah, The optical efficiency of three

10 different geometries of a small scale cavity receiver for concentrated solar applications,

11 Appl Energ 179 (2016) 1081-1096.

12 [21] A.B. Zavoico, Solar Power Tower Design Basis Document, Sandia National

13 Laboratories, Albuquerque, New Mexico, USA, 2001.

14 [22] P.K. Falcone, A handbook for solar central receiver design, Sandia National

15 Laboratories, Livermore, California, USA, 1986.

16 [23] F.P. Incropera, T.L. Bergman, A.S. Lavine, D.P. DeWitt, Fundamentals of heat

17 and mass transfer, 7th edition ed., JOHN WILEY & SONS, INC., Hoboken, New Jersey,

18 USA, 2011.

19 [24] B.S. Petukhov, Heat Transfer and Friction in Turbulent Pipe Flow with Variable

20 Physical Properties, Advances in Heat Transfer 6 (1970) 503-564.

21 [25] D.L. Siebers, J.S. Kraabel, Estimating convective energy losses from solar central

22 receivers, Sandia National Laboratories, Albuquerque, New Mexico, USA, 1984.

37 / 38
ACCEPTED MANUSCRIPT

1 [26] D.L. Siebers, R.F. Moffatt, R.G. Schwind, Experimental, Variable Properties

2 Natural Convection From a Large, Vertical, Flat Surface, Journal of Heat Transfer

3 107(1) (1985) 124-132.

4 [27] E. Achenbach, The effect of surface roughness on the heat transfer from a circular

5 cylinder to the cross flow of air, Int J Heat Mass Tran 20(4) (1977) 359-369.

6 [28] C.-J.Winter, R.L. Sizmann, L. L.Vant-Hull, Solar Power Plants: Fundamentals,

7 Technology, Systems, Economics, Springer-Verlag, Berlin, Germany, 1991.

8 [29] A. Sánchez-González, M.R. Rodríguez-Sánchez, D. Santana, Aiming strategy

9 model based on allowable flux densities for molten salt central receivers, Sol Energy

10 (2016).

11 [30] H. Benoit, L. Spreafico, D. Gauthier, G. Flamant, Review of heat transfer fluids in

12 tube-receivers used in concentrating solar thermal systems: Properties and heat transfer

13 coefficients, Renewable and Sustainable Energy Reviews 55 (2016) 298-315.

14

38 / 38
ACCEPTED MANUSCRIPT

Highlights
>3D transient model was built to analyze transient behaviors of solar receivers.

>Flowrate variation affects markedly inner surface temperatures near the outlet.

>30% reduction in the flowrate causes the receiver to be 630 °C in 3 s.

>Radial temperature difference is larger at higher DNI even with lower flowrate.

Anda mungkin juga menyukai