Vous êtes sur la page 1sur 10

Applied Energy 123 (2014) 232–241

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Structural reliability analysis of parabolic trough receivers


Zhiyong Wu a,⇑, Dongqiang Lei a, Guofeng Yuan a, Jiajia Shao b, Yunting Zhang b, Zhifeng Wang a
a
Key Laboratory of Solar Thermal Energy and Photovoltaic System, Institute of Electrical Engineering, Chinese Academy of Sciences, No. 6 Beiertiao, Zhongguancun, Beijing 100190,
PR China
b
Himin Solar Co., Ltd., Dezhou, Shandong 253090, PR China

h i g h l i g h t s

 A method combined with MCRT, CFD and FEA is developed for studying for PTR’s structural reliability.
 The effects of HTF velocity and temperature on the temperature difference of PTR were investigated.
 The effects of HTF velocity and temperature on the thermal stress of PTR were investigated.
 The metal tube deflection is weakly dependent on the thermal stress from non-uniform heat flux.
 The metal tube deflection is highly dependent on its supporting conditions and neighboring PTR’s state.

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

Article history: Parabolic trough receivers typically account for 30% of the cost of a solar field. Furthermore, experiences
Received 22 August 2013 show the cost of operation and maintenance of parabolic trough receivers over the lifetime of the system
Received in revised form 14 February 2014 can be quite high as regular maintenance is required. Therefore, the receiver’s cost and structural reliabil-
Accepted 17 February 2014
ity is crucial to guarantee the parabolic trough plants work steadily, safely and above all, economically.
Available online 18 March 2014
The structure of a parabolic trough receiver is principally a stainless steel inner tube surrounded by a
glass outer tube. Because the stainless steel tube has 4 m long and is easy to get bent, the glass outer tube
Keywords:
is quite fragile and therefore prone to break. Experiences in field operations show the bending of stainless
Parabolic trough receiver
Coupled heat transfer modeling
steel tube is the main reason of parabolic trough receiver’s structural failure. This study mainly investi-
Thermal stress gates why and how the stainless steel tubes get bent. In this paper, an indoor experiment, numerical
Field measurements simulations (CFD and FEA) and field measurements are conducted to jointly investigate the causes of
Structural reliability stainless steel tube’s bending. An indoor experiment is conducted on a heat loss test-rig, and the deflec-
tion of the stainless steel tube under different isothermal conditions is studied. Numerical simulations are
used to investigate the detailed temperature distribution, and its corresponding structural deformation of
the stainless steel tube of parabolic trough receivers. The simulations adapt a MCRT code, FLUENT soft-
ware and ANSYS workbench. Finally, field measurements are conducted on a 36 m long parabolic trough
collector to investigate the stainless steel tube bending under different working conditions. This study
shows that improper installation and operational practices of parabolic trough receivers are the main
causes for the structural failure of the parabolic trough’s receiver. This study is of great importance to
guide the installation and operation of parabolic trough receivers.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction capability with other fuels to meet the electricity demand at night.
This means that CSP technology is very promising and is presently
The power from CSP (Concentrated Solar Power) is more stable the only renewable energy which can be utilized as the base-load
than that from the PV (Photovoltaic) technology because this power for electricity grid.
system has a large heat storage capacity. In addition, the CSP plants It has been more than 28 years since the first nine large
integrated with thermal storages can deliver electricity during commercial-scale parabolic trough power plants were constructed
cloud cover or after sunset and has a good hybrid integration in 1984 by Luz International Limited in the Mojave Desert,
California [1]. Unfortunately, these power plants experienced
⇑ Corresponding author. Tel.: +86 10 82547050; fax: +86 10 62587946. unacceptable high failure rate of the PTR (Parabolic trough
E-mail address: wuzhiyongspecial@gmail.com (Z. Wu). receiver) during the first few years [2]. According to the reported

http://dx.doi.org/10.1016/j.apenergy.2014.02.068
0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
Z. Wu et al. / Applied Energy 123 (2014) 232–241 233

Nomenclature

Cp specific heat (J/(kg K)) ag absorptance of glass envelope


Cl k  e model constants k thermal conductivity (W/(m2K))
D elasticity matrix kin turbulent kinetic energy at the inlet
D1 glass envelope diameter (m) m kinematic viscosity
DNI direct normal irradiance (W/m2) mp Poisson’s ratio
g acceleration due to gravity (m/s2) e strain tensor
0
Ku Knudson number (l/L) e equivalent strain
L characteristic length (m) ein turbulent dissipation rate at the inlet
l molecular mean free path (m) r0
stress tensor
LPTR PTR Length r equivalent stress
LG glass envelope length q density (kg m3)
Nu Nusselt number (hL/k) l dynamic viscosity (kg m1 s1)
P pressure (Pa) lt turbulent viscosity (Cpqk2/e)
Pk pressure (Pa) m kinematic viscosity (m2 s1)
Pr Prandtl number (Cpl/k) es emissivity of the selective coating
Q heat flux
Ra Rayleigh number (Grx Pr ) Subscripts
Re Reynolds number (qud/l) a absorber
T temperature (K) x coordinate X
T1 thickness of absorber tube (m) f heat transfer fluid
T2 thickness of glass envelope (m) a absorber tube
U velocity (m s1) in inlet
dif difference
Greek symbols max maximum
a thermal expansion coefficient min minimum
a1 thermal diffusivity

data, the average annual PTR replacement rate was still 5.5% for [27]. The fluid–solid coupled heat transfer was solved with the
SEGS III–VII during the period 1997–2001 [2]. The most recent data finite volume method. The concentrated solar heat flux was used
for SEGS indicated that annual receiver tube failures had decreased as a boundary condition of the fluid–solid coupled heat transfer
to 3.37% of the total field receivers [3]. Since 2000, extensive works modeling. In the annular vacuum gap, the conduction was ignored,
have been done on the PTR by the parabolic trough industry and and the infrared radiation heat transfer was treated as a gray
solar research laboratories around the world. These studies focus enclosure and modeled with S2S radiation model [25]. In the CFD
on the glass-to-metal seal [4–6], selective coating [7–12], thermal simulations, the studied PTR (see Fig. 1) was assumed to be struc-
performances [13–17], vacuum reliability [18,19] and heat transfer turally rigid, and its surfaces have wavelength independent prop-
enhancement of the stainless steel tube (also named absorber erties. The HTF (heat transfer fluid) had temperature dependent
tube) [20–22]. Specifically, Wang [23] and Shuai [24] adopted properties. The temperature field was used to investigate the ther-
MCRT (Monte-Carlo ray tracing) method, CFD (Computational mal stress and the corresponding strain. Finally, field measure-
Fluid Dynamics) and FEA (Finite Element Analysis) codes to ana- ments of the PTR are done on a 36 m long parabolic trough
lyze the temperature distributions, thermal stress of eccentric tube collector. The field measurements include the alignment of sup-
receivers. However, the structural reliability of PTR has not been porting arms, installation and monitoring the bending of stainless
thoroughly studied according to the authors’ best knowledge. Sun- steel tube under different working conditions. This results of this
Lab projections indicate the replacement rates should decrease to study show that improper installation and operational practices
0.5% till 2020 [2]. This shows there is still much room for improve- of PTR are the two main items which cause the PTR’s structural
ment of the PTR. failure. The authors believe this study is of great importance to
This study mainly investigates why and how the stainless steel guide the design, optimization, installation and operation of PTR,
tubes get bent, because field operation experiences show this is and to the CSP industry.
likely to break the glass envelope, and thus cause the structural
failure of PTR. In this paper, an indoor experiment, numerical sim-
ulations (CFD and FEA) and field measurements are conducted to 2. Numerical simulations
jointly investigate the bending of PTR’s stainless steel tube. The in-
door experiment is to investigate the deflection (Note: the deflec- The simulations consist of Monte-Carlo ray tracing, CFD and FEA
tion refers to the width changing of annular gap because of the modeling. Firstly, Monte-Carlo ray tracing was used to calculate
stainless steel tube bending under different conditions. In addition, the concentrated solar radiation heat flux distribution on the stain-
the authors ignored the deformation of glass envelope in this less steel tube (absorber tube) of PTR. This flux distribution is trea-
study) of stainless steel tube of PTR under different temperature ted as a boundary condition of the CFD analysis. Secondly, CFD was
conditions. Numerical simulations are used to investigate the de- used to model the flow and heat transfer (including convection,
tailed temperature distribution, and its corresponding structural conduction and radiation heat transfer) in the whole PTR. Through
deformation of the stainless steel tube of PTR. The simulations CFD simulation, the detailed temperature distribution of the whole
are conducted by combining a MCRT code, FLUENT [25] software PTR can be obtained. Finally, the temperature distribution was
and ANSYS workbench [26]. The non-uniform concentrated solar used as a temperature load of PTR in the FEA analysis. Together
heat flux on the steel pipe was obtained through the MCRT method with the gravity load and supporting boundary conditions, the
234 Z. Wu et al. / Applied Energy 123 (2014) 232–241

Fig. 1. The structure of PTR.

stress and strain field in PTR can be obtained. By varying the con- Glass
centrated heat flux distribution, the flow speed of HTF and adjust- envelope
ing the supporting boundary conditions, sensitivity studies are
Vacuum
conducted so as to study the bending of stainless steel tube under
annular gap
various working conditions, and thus verify the causes of the Qsolar
breakage of the glass envelope.
Specially, the Monte-Carlo ray tracing for PTR has been vali- Rconvection
dated in [28]. The CFD modeling of the whole PTR structure has
also been validated in [29], and the CFD results agree with the
Qloss
experimental data within 5% of deviation. Although the FEA analy- Rradiation
sis has not been validated, the maximum working temperature of Heat transfer
PTR is about 700 K (because the maximum working temperature fluid
Qobtain
of HTF is 670 K) and thus the stainless steel tube works in the elas- Rconvection Rconduction
tic range of this material. Within the elastic range, the relationship
of strain and stress is linear, and thus the accuracy of FEA analysis Rradiation
lies mostly on the temperature load, which is directly from the CFD Absorber
results. Therefore, the authors consider the methodology of numer-
ical simulations in this study is reliable.
(Ambient air, Sky)

2.1. Monte-Carlo ray tracing (MCRT) Fig. 2. Cross-sectional schematic of parabolic receiver tube with thermal resis-
tances and heat flows.
MCRT has been successfully used in studying the heat transfer
characteristics of PTR [27,30,31], and the numerical results agree
well with experimental data. In fact, the MCRT work only provides away by the HTF through convection. In the annular vacuum gap,
a non-uniform heat flux distribution. The MCRT is a well known the selective coating exchanges heat with the inner surface of glass
method, and is widely used in the optical simulation of CSP. The envelope through convection and radiation. At the outer surface of
consensus of the heat flux on the PTR is a double-peak distribution. the glass envelope, a quantity of thermal energy is released to the
In this study, a double-peak heat flux distribution was obtained environment through convection and radiation. A cross-section of
through an in-house MCRT code. the PTR with thermal resistances and heat flows are shown in
In addition, the authors noticed that the real flux distribution on Fig. 2.
the absorber tube is dependent on the direct normal irradiance, Within the structure of PTR, the HTF flow, radiation, convection
tracking accuracy, concentrator’s structural accuracy and the shape and conduction heat transfer are interdependent. In this study, this
of absorber tube (the tube is not strictly straight in reality), but this interdependent physical problem is jointly simulated with FLUENT
is outside the scope of this study. software.

2.2. CFD modeling 2.2.1. Flow, conduction and convection heat transfer modeling
The HTF flow in PTRs is essentially flow in a straight pipe. The
The general structure of solar receiver is a stainless steel inner flow and heat transfer inside a pipe with constant wall heat flux
tube surrounded by a glass outer tube, and between the two tubes or constant wall temperature have been thoroughly studied, even
is an annular vacuum gap, see Fig. 1. On the outer surface of the analytical solutions exist [32] and heat transfer correlation was
stainless steel tube is a selective coating which has wavelength proposed since 1942 [33]. However, the studies of flow and heat
dependent properties of high absorptance to the radiation of solar transfer in a pipe under non-uniform solar radiation flux are rare.
spectrum, and low emittance to the infrared spectrum. In addition, the bending of stainless steel tube of PTR is directly re-
The concentrated solar radiation flux is transmitted through the lated to the temperature distribution. Therefore, it is necessary to
glass envelope, and falls on the absorber tube. The absorber tube investigate the detailed temperature field of PTR. In this study,
absorbs the concentrated radiation flux and converts it into heat. the flow, conduction, radiation and convection heat transfer within
The heat is then conducted to the inner surface, and then taken the whole PTR are jointly modeled through FLUENT software.
Z. Wu et al. / Applied Energy 123 (2014) 232–241 235

The Reynolds number (Re = qfud/l) of HTF flow in the absorber transfer from the glass envelope’s outer surface to the environment
tube is approximately 2  105, and the flow is turbulent. Therefore, was considered by using a radiative boundary condition.
a turbulence model (standard k-epsilon model [25]) is used in this
study. 2.2.3. Boundary conditions
The governing equations are based on the three dimensional
steady Reynolds-averaged Navier–Stokes equations, which consist (1) For the fluid domain, the inlet and outlet boundary condi-
of the conservation equations of mass, momentum and energy tions are defined as:
[34]. In this study, the temperature of HTF ranged from ambient c q k2
Inlet: u = uin, T = Tin, kin ¼ 0:005qu2x , ein ¼ l lf in
t
temperature to 670 K. Thus, temperature independent properties where cl = 0.09, lt = 100, and at the inlet ux = uin.
of the HTF were considered in this study [35]. Outlet: fully-developed conditions [40].
The conduction heat transfer in PTR refers to the heat transfer in The real HTF temperature in a working PTR ranges from about
the solid of stainless steel tube and glass envelope. The heat con- 370 K to 670 K, and in this study, Tin = 300–600 K and uin = 0.25–
duction in the solid is governed by a heat conduction equation, 2.5 m/s.
which was solved directly in the simulation. The fluid–solid inter-
faces were defined as walls with coupled heat transfer conditions. (1) The selective coating is defined as a wall with a distributed
The heat transfer in the vacuum annular gap is related to the heat flux. The emissivity of the selective coating was defined
rarefied gas [36,37] in which the Knudson number (Ku = l/L, the ra- as a temperature dependent value (see Table 1).
tio of Molecular mean free path to the characteristic length) is used
to verify the heat transfer mode. In a typical PTR, the Knudson A circumferential distributed energy source was defined on the
number is larger than 10, this means the molecular mean free path glass envelope, which was calculated from the absorptance value
is so large compared with the characteristic length that the colli- and the concentrated solar flux distribution from MCRT code.
sions between molecules can be ignored. Therefore, the heat trans- Although the glass envelope has a low absorptance to the solar
fer mode is conduction dominated. But results from Roesle [38,39] radiation, the authors believe that it is considerable to the glass
and Sharipov [37] show the heat conduction through vacuum envelope’s temperature distribution because the heat flux is con-
annular gap of PTR is less than 2.5 W/m when Ta = 670 K. Com- centrated. Fig. 3 presents the heat flux on the PTR’s selective coating
pared with the total heat loss of PTR, heat losses of 2.5 W/m across and glass envelope. The computed magnitude and location of the
the vacuum annular gap is so small that it was ignored in this peaks of the flux agrees well with the distribution in [31]. The effect
study. of PTR’s shading on the flux distribution was not considered, and
thus the flux distribution is more flat near angle 0° (see Fig. 3).
2.2.2. Radiation heat transfer modeling
The solar radiation and the radiation heat transfer in the PTR are (1) The outer surface of the glass envelope was defined with a
treated separately because their radiations have different wave mixed boundary with convection and radiation. The
length, on which the PTR’s optical properties are dependent. Con- Stefan–Boltzmann law is used to calculate the net radiation
cerning the solar radiation, the glass envelope has a transmittance transfer by assuming that the glass cover is a small convex
of 0.96, an absorptance of 0.02, and the selective coating has an gray object in a large blackbody cavity (sky).
absorptance of 0.95. Therefore, the calculated double-peak heat
flux distribution from MCRT code was defined as the heat flux The convection loss is assumed as natural convection and the
boundary condition to the stainless steel tube. The glass envelope’s heat transfer coefficient was estimated through the correlation be-
absorption of solar radiation was considered so as to predict its low from literature [41].
temperature with more accuracy. ( )2
The infrared radiation heat transfer in the vacuum annual gap hD1 0:387R1=6
a
Nu ¼ ¼ 0:6 þ ð1Þ
was simulated with surface-to-surface (S2S) radiation model, the k 8=27
½1 þ ð0:559=PrÞ9=16 
participating surfaces were assumed to be gray and diffuse. The
g aDTD31
glass envelope is assumed to be opaque to the radiation within where Rayleigh number Ra ¼ a1 m and Prandtl number Pr ¼ mk.
the infrared energy spectrum. The selective coating has a temper-
ature independent radiative property [16], and the optical proper- (1) The inner surface of the absorber tube were defined as cou-
ties of the other surfaces are listed in Table 1. Radiation heat pled fluid–solid interfaces, in which the energy is conserved.

Table 1
Specifications of the studied parabolic trough receiver.

Quantity Description Unit Value


LPTR PTR length m 4.060
LG Glass envelope length m 3.934
D1 Absorber tube diameter m 0.07
T1 Thickness of absorber tube m 0.003
D2 Glass envelope diameter m 0.12
eg Emissivity of glass envelope % 89
T2 Thickness of glass envelope m 0.003
a Absorptance of the selective coating (within the solar energy spectrum) % 95
es Emissivity of the selective coating % 10.9  2.98  102T + 6.4  105T2
p0 Absolute pressure in annular vacuum gap Pa <0.01
ag Absorptance of glass envelope – 0.02
k Thermal conductivity Wm1 K-1 25
vp Poisson ratio – 0.25
E Young’s modulus (GPa) GPa 220
a Thermal expansion coefficient 106 K1 17.2
236 Z. Wu et al. / Applied Energy 123 (2014) 232–241

50000 DNI: 1000W/m


2 The stress fields are the combined effect of the thermal gradient
o
Angle incidence: 90
and gravity. The thermal stress can be calculated as:
Absorber tube
40000 r¼De ð2Þ
Glass envelope
where r is stress, D is the elasticity matrix, and e is the strain. The
Heat flux (W/m2)

30000 Glass source


studied material is assumed to be isotropic and within the regime of
elastic deformation, the stress–strain relations can be written in
20000 Cartesian coordinates as:
1
10000 ex ¼ ½rx  v p ðry þ rz Þ þ arTðx; y; zÞ ð3Þ
E

0 1
ey ¼ ½ry  v p ðrz þ rx Þ þ arTðx; y; zÞ ð4Þ
-180 -120 -60 0 60 120 180
E
Angle (o)
1
ez ¼ ½rz  v p ðrx þ ry Þ þ arTðx; y; zÞ ð5Þ
Fig. 3. Heat flux distributions on parabolic trough receiver (DNI = 1000 W/m2). E
where E, vp and a are the modulus of elasticity, Poisson’s ratio and
(2) The heat losses from the support structures are ignored
the coefficient of thermal expansion, respectively. In the above
because Forristall [15] estimated that it accounts for about
equation, DT(x,y,z) represents the temperature gradient at a point
1–4% of the total system heat losses. In addition, the outer
(x,y,z). The temperature fields were directly interpolated by the
surfaces of the bellows were defined as adiabatic walls
CFD results.
because they are usually wrapped with insulation materials.
The principal stresses are calculated by solving the Eigen values
(r1, r2, r3) of a three order stress tensor, and then the von-Mises
2.2.4. Numerical method equivalent stress is computed from,
A three dimensional HEXA mesh was generated and used in this rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
study. Dense grids were generated near the bellows where large 1
r0 ¼ ½ðr1  r2 Þ2 þ ðr2  r3 Þ2 þ ðr3  r1 Þ2  ð6Þ
temperature gradients exist. This three dimensional conjugate heat 2
transfer problem was solved using the CFD software FLUENT. The 0
The equivalent stress r is related to the equivalent strain e0
standard k  e model (high Reynolds number) was used along with through r0 = Ee0 .
the standard wall function [25] in this study.
The SIMPLE algorithm was used to couple the pressure and the 2.3.2. Loads and boundary conditions
velocities with the second order upwind method used for the The main load on a working PTR is the temperature, because the
advection terms in the momentum equations and for the energy non-uniform heat flux on the stainless steel tube brings huge tem-
equation. User defined functions (UDF) were used to define the perature gradients, which causes considerable thermal stresses.
concentrated solar radiation flux on the selective coating and the The temperature fields obtained from CFD simulations are directly
circumferential distributed source term of the glass envelope. used as the temperature conditions in the FEA simulations.
To guarantee that the computed results were accurate and reli- The supporting and displacement conditions (see Fig. 4) are de-
able, the used discretization of all solved equations was second or- fined as follows:
der, and the convergence criterion was set as 2  105 for all the
equations. In addition, a grid independence study was also con- (1) For the simulation of a single PTR, the two ends were defined
ducted. Detailed information of the used CFD methodology can as free-supporting conditions.
be found in [29]. (2) For the simulation of 9 PTRs, small surfaces were cut at the
locations of supporting arms and their displacement were
2.3. FEA modeling defined as free along the PTR’s axial direction, and 0 along
other directions.
The PTR is generally manufactured as segments of 4 m long.
Individual receiver segments are end-to-end welded together and Disy ¼ Disz ¼ 0; Disx ¼ free:
then supported at the focal line of the parabolic trough collector In addition, the gravity of the PTR was considered. Although the
with brackets. To accommodate the axial deformation of PTRs at PTR rotate with the whole parabolic trough collector, the direction
working state, the bracket is specially designed and can slightly ro- of the gravity was assumed as fixed and coincides with the nega-
tate. The PTR’s structure is also special because it has bellows at the tive Z coordinate.
ends. The bellow has a small flexural and compressive strength,
and is pre-stressed when the PTR is manufactured. This means
the bellow can accommodate a certain amount of axial deforma-
tion and bend of the stainless steel tube. Above all, the actual axial (a) S1 S2
Z HTF flow
direction gravity
deformation and deflection of the stainless steel tube under the
S1: dx=dy=0, dz=free
non-uniform thermal stress and gravity are small, or else the glass Y X
S2: dx=dy=0, dz=free
envelope will break. Therefore, the authors assumed the bellows
and glass envelope have no effect on the deformation of the stain-
less steel tube, and thus only the strength of the stainless steel (b) S9
S1 S2 S3 S0 S4 S5 S6 S7 S8
tubes was analyzed in the FEA simulations.
S0: dx=dy=dz=0
S1,…,9: dx=dy=0, dz=free
2.3.1. Constitutive relations
The FEA was used to calculate the structural deformation of Fig. 4. Schematic illustration of the coordinate system and boundary conditions of
PTR. To calculate the deformation, the stress field was computed. FEA simulation. (a) Single PTR, (b) 9 PTRs.
Z. Wu et al. / Applied Energy 123 (2014) 232–241 237

The authors assume the bellows and glass envelope have no ef- 3.2. Field measurement
fects on the stainless steel tube’s strength.
The bending phenomenon of stainless steel tube was also stud-
ied in a 36 m long test loop (see Fig. 6) at Yanqing, Beijing. This
3. Experimental studies
facility was built for demonstrating different components and
assemblies of parabolic trough collectors and has been operated
3.1. Indoor measurement
jointly by IEE,CAS and Himin Solar Co. since 2010. As depicted in
Fig. 6, the loop consists of 3  12 m long collector modules. Cool-
Two years of field operation experiences of Himin’s PTR70 in
ing, pumping and measuring systems (pressure, temperature and
Yanqing, Beijing of China, show the most likely cause of the struc-
volumetric flowrate) are located on this loop together. There are
tural failure of PTR is the bending of stainless steel inner tube.
no heating systems in this loop. The pump has a lift of 30 m, and
However, the reasons of the bending of stainless steel tube are
can provide mass flowrates of up to 8 kg/s. A transducer is used
not clear. During the heat loss testing of PTR, the technicians found
to control the rotating speed of pump and then adjust the mass
the stainless steel tubes are considerable bent. Therefore, the heat
flowrate. Mass flow measurements are realized using a vortex
loss testing was adjusted to verify the degree of the stainless steel
flowmeter (DY040-NALSR4-4N/HT2, YOKOGAWA). 7 Pt100 ther-
tube bending under different isothermal conditions.
mocouples immersed in the HTF are used for temperature mea-
Fig. 5 shows the heat loss test-rig, on which a PTR (4.06 m long)
surements (note: 1 locates at the inlet, 3 locate at the end of 6th
was installed. During the experiment, the PTR was heated by two
PTR and 3 locate at the outlet. In addition, some thermocouples
electric cartridge heaters, which was surrounded by copper pipes
were used for studying the flow stratification [42] and the mea-
(50 mm outer diameter) and inserted into the PTR from both ends.
sured data are not presented in this study). In order to eliminate
The copper pipes were used to guarantee the temperature unifor-
factors (sensor and transmitter drift and non-linearities, limited
mity at both the longitudinal and circumferential directions. Three
sensor calibration precision) that typically affect the quality of
bolt heads protruded from each end of the copper pipe center of
small temperature difference data at high temperatures, the sen-
the receiver to prevent it from touching the inner absorber surface.
sors are calibrated relative to one another. In doing so, an agree-
A proportional-integral-derivative control system was used during
ment of the sensors and their entire measurement chain to about
the heat loss testing to adjust heater power and absorber temper-
1.0 K is obtained. Because this experiment is to investigate the
atures until all the absorber temperatures approached the setting
deflection stainless steel tube under different working conditions,
values. In order to decrease the influence of wind speed, the test-
particular care has been taken with regards to the installation
rig was placed indoor and away from the window and door during
and alignment of the 9 PTRs at the beginning of the experiment.
the testing period. With the self-adapting PID control program, the
To monitor the deflection of the stainless steel tube next to the
PTR can reach a thermal equilibrium state after 3 h. Thermocouples
glass envelope, one digital SLR camera (Nikon D700) was fixed at
indications show that the temperature deviations along the longi-
the edge of the concentrator and set up to take photos at every
tudinal direction were less than 6 K when the stainless steel tube
15 s. The direction of camera is always vertical to the axis of stud-
was at a thermal equilibrium state of 670 K [16].
ied PTR. In addition, a metal ring was fixed on the PTR, which acted
The degree of the stainless steel tube bending was evaluated by
as a reference in the image process. Both the camera and PTR are
measuring the width of the annular vacuum gap of PTR under var-
fixed on the concentrator which rotates together to track the sun.
ious temperature conditions. Firstly the studied PTR was heated to
The deflections of stainless steel tube can be directly measured
a target temperature through the cartridge heaters and controlled
in computer by rooming out the images, because the diameter of
by a self-adapting PID program. When the PTR reached a thermal
the tube is fixed and the border of the tube can be easily identified.
equilibrium state [16], then quickly removed the two cartridge
heaters and as well as simultaneously measured the gap width
with a ruler. The width of the annular vacuum gap at different cir-
4. Results and discussion
cumferential directions was also measured by rotating the PTR. In
addition, the two ends of the PTR were always kept freely-sup-
4.1. Temperature gradient of PTR
ported during this measurement.
Because it is not easy to accurately measure the annular gap
The thermal performance of PTR has been widely investigated
width, as a result the uncertainty of this measurement was large
[14,28,30,43,44]. The temperature distribution of PTR’s have also
and estimated at approximately about 1 mm. In this study, 12 PTRs
been investigated in detail by the author [29]. However, to the best
were tested at an ambient temperature of 670 K.
of the authors’ knowledge, there is very little literature investigat-
ing the relationship between the temperature gradient in the stain-
less steel tube and the PTR’s working conditions. Fig. 7 shows the
temperature distribution of the absorber tube and glass envelope
on a sectional plane. The temperature gradient in the stainless steel
tube was evaluated by the temperature difference on a sectional
plane which was calculated by Tdif = Tmax  Tmin. Studies have
shown that the higher Tdif, the larger thermal stress in the stainless
steel tube, and thus the larger deflection of the stainless steel tube.
Parametric studies of HTF velocity and temperature on the temper-
ature gradient were conducted in this paper.
Fig. 8 presents the relationship between Tdif and the HTF veloc-
ity and temperature. Fig. 8 shows the higher the HTF velocity, the
lower Tdif is, and the higher the inlet temperature of HTF, the lower
Tdif is. This means that the PTR works under a higher risk when the
HTF velocity is low or the HTF temperature is low, or both the
velocity and temperature are low. A high HTF velocity is advanta-
Fig. 5. PTR heat loss test stand at IEECAS. geous to the heat transfer and the PTR’s security. However, a big
238 Z. Wu et al. / Applied Energy 123 (2014) 232–241

Fig. 6. Picture of the test facility in Yanqing, Beijing.

0.04
(a) 659.286 120
HTF temperature

Temperature difference (C)


300 K
659.286 100 400 K
Temperature (K) 500 K
0.02
715 600 K
710.357 80
705.714
66

701.071
8.
57

696.429
1

60
Z (m)

0 691.786
3

687.143
14
7.

682.5
68

68

40
7.1

677.857
43

673.214
668.571
-0.02 663.929 20
69

659.286
6.

654.643
42

691.786
9

650
691.786
0
0 0.5 1 1.5 2 2.5 3
-0.04
Velocity (m/s)
-0.04 -0.02 0 0.02 0.04 0.06
Fig. 8. Relationship between temperature differences and HTF velocity and
X (m) temperature.

0.06
(b) 337.143 single PTR show that the deflection of the stainless steel tube is
nearly a constant of 2.5 mm. This illustrates a uniform temperature
337.143
0.04 337.143 Temperature (K) load of less than 670 K has no considerable effects on the bending
380
375.714
of stainless steel tube. However, the technicians experienced a
0.02 371.429
367.143
deflection of about 10 mm during the heat loss experiments.
4
71
34
5. 362.857 Therefore, the heat loss test-rig was adjusted, and then an indoor
Z (m)

35
0 358.571
0 354.286 measurement of the stainless steel tube’s deflection was done un-
350
345.714 der isothermal conditions of Ta which varies from ambient temper-
341.429
1

-0.02 ature to 670 K. The test-rig and measuring procedure were


.57

337.143
35 8

332.857
328.571
described at Section 3.1. The results show the deflections of the
-0.04 367.143 324.286
320
stainless steel tube ranges from 3 to 6 mm when the PTR is under
362.857 different temperature conditions. By measuring the stainless steel
-0.06 362.857 tube’s deflection at different circumferential directions, the techni-
cians also found the deflection of stainless steel tube larger than
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
3.5 mm is not because the temperature is high, but because the
X (m)
stainless steel tube itself or the glass tube is not strictly straight.
Fig. 7. Temperature distribution of PTR at constant Y. (a) Absorber tube, (b) glass This conclusion proves the validity of FEA simulation, and indicates
envelope. that an isothermal temperature (Ta < 670 K) has no considerable ef-
fects on the stainless steel tube bending.

pump is required, and accordingly the pressure loss will be high


and the temperature increment of HTF through a collector will 4.2.2. FEA simulation of PTR
be low. These are not in accordance with the system design, and Firstly, a single PTR under different temperature conditions was
are sometimes uneconomical. Above all, the low temperature of numerically simulated, with free-supporting conditions at two
the HTF is sometimes unavoidable. For example, the temperature ends. Fig. 9 presents the relationship between the stainless steel
of the HTF is at an inherent low at the start-up period. tube deflection and the HTF velocity and temperature (In the fig-
ure, the negative value indicates the deformation towards the
4.2. Stainless steel tube deflection coordinate negative Z). This figure shows that the higher the HTF
velocity, the lower the stainless steel tube deflection; the lower
4.2.1. Indoor measurement of the deflection of a single PTR the inlet temperature of the HTF, the lower the stainless steel tube
When the absorber temperature varies from ambient tempera- deflection. The FEA results coincide with the temperature differ-
ture to 670 K and the load has only the gravity, the FEA results of a ence on a sectional plane of the stainless steel tube from the CFD
Z. Wu et al. / Applied Energy 123 (2014) 232–241 239

Deflection (mm) (Z axis)


0

-10
-20
Deflection (mm)

-20 -40 (a)


-2.03 0 2.03
-30 HTF temperature
X (m)
300 K
400 K
2
-40 500 K
600 K
(b)
0

Deflection (mm) (Z axis)


-50
0 0.5 1 1.5 2 2.5 3 -2
Velocity (m/s)
-4
Fig. 9. Relationship between the maximum directional deformation of stainless
steel tube and HTF velocity and temperature (single PTR). -6

-8
simulation. The higher the temperature difference that exists, the
-10
larger the stainless steel tube deflection that will appear. The com-
puted FEA results clearly illustrate the trend of deflection changing
-12
with HTF velocity and temperature. However, the authors notice -16.24 -12.18 -8.12 -4.06 0 4.06 8.12 12.18 16.24 20.3
the predicted deflection is too large and unreasonable. The authors X (m)
contribute the large deflections to the used free-supporting condi-
Fig. 11. Computed directional deformation of the axis of studied PTR(s). (a) Single
tion, which is actually not suitable here.
PTR; (b) 9 PTRs.
To avoid unreasonable results, FEA simulations of 9 PTRs (the
reason for simulating 9 PTRs is because the test loop is 36 m long,
on which 9 PTRs are installed) under different working conditions
were also conducted. Fig. 10 presents the relationship between the PTR’s state. Fig. 11 presents the computed directional deformation
stainless steel tube deflection and the HTF velocity and tempera- of 1 PTR and 9 PTRs. In Fig. 11(b), the black dot refers the fixed sup-
ture. This figure also verifies the relationship between the stainless port (see Figs. 4 and 6). Fig. 11 clearly verifies the conclusion that
steel tube deflection and HTF velocity and temperature. The simu- ‘‘the stainless steel tube deflection is weakly dependent on the
lation results also show that the deflections of 2 PTRs at the two thermal stress, and highly dependent on its supporting conditions
sides are much larger than that of the middle 7 PTRs. The maxi- and neighboring PTR’s state’’.
mum deflection of the PTRs at the two sides reaches 16 mm. How-
ever, the magnitude of deflection of the PTRs in the middle is 4.2.3. Field measurement
always less than 4 mm, even when the HTF velocity is decreased When studying the PTR’s behavior under different working con-
to 0.25 m/s. The authors also contribute the large deflections of ditions, the 36 m long loop in Yanqing, Beijing of China (see Fig. 6)
the PTRs on both sides of PTR to the boundary conditions defined is not an ideal test-rig because there is no heating system, and the
on them. capacity of cooling system (which was not used in this test) and
Therefore, the stainless steel tube bending depends not only on HTF tank are small. The HTF cycles directly in the loop, and then
the HTF velocity and temperature, but also on the defined support- the HTF temperature steadily increases (see the temperature vari-
ing conditions. In fact, the FEA results shows that the stainless steel ation curves in Fig. 12). This means the operators cannot control
tube deflection is weakly dependent on the thermal stress, and the inlet HTF temperature well during the test. The HTF mass flow-
highly dependent on its supporting conditions and the neighboring rate can be adjusted with a valve. The HTF temperature, mass flow-
rate and the behavior of the PTR are continuously monitored in the
test. During the test, the HTF temperature varied from ambient
temperature to 640 K (for security reason, the test was stopped
10 when the maximum HTF temperature reached 640 K) and the mass
flowrate varied from 3 to 18 m3/h, which have covered the typical
5 working conditions of the studied PTR. Therefore, the authors be-
lieve that this loop is capable of testing the PTR.
Deflection (mm)

0 Fig. 12 presents the variation curves of HTF temperature and


flowrate in the field measurement. In Fig. 12(a), the HTF tempera-
-5 ture changes from 320 K to 580 K, and the flowrate changes from
18 m3/h to 8 m3/h. In Fig. 12(b), the HTF temperature changes from
-10 300K,1.0m/s
300K,2.0m/s
490 K to 620 K, and the flowrate changes from 11 m3/h to 6 m3/h.
300K,3.0m/s Particularly, the test of Fig. 12(a) was undertaken between 11:40
600K,1.0m/s
-15 600K,2.0m/s AM and 12:40 PM. During this period, the concentrator was nearly
600K,3.0m/s
facing up, and the direction of gravity approximately overlaps with
-20 coordinate, negative Z.
0 1 2 3 4 5 6 7 8 9 10
Fig. 13 presents 5 figures of one PTR (the second one from the
No of the studied PTR
HTF inlet) during the experiment. The images #1, #2 were taken
Fig. 10. Relationship between the maximum directional deflection of stainless steel at noon while the concentrator was facing up. Theoretically, the
tube and HTF velocity and temperature (9 PTRs). stainless steel tube will have the largest deflection. However, the
240 Z. Wu et al. / Applied Energy 123 (2014) 232–241

600 HTF flowrate

HTF inlet temperature


500
T (K) / Flowrate (*0.1 m/h)

HTF outlet temperature

DNI

400

300 1200

DNI (W/m2)
200 800

100 400
(a)
#1 #2
0
0 10 20 30 40 50 60
Time (min)

600 (b)
T (K) / Flowrate (*0.1 m/h)

500
HTF flowrate

HTF inlet temperature


400
HTF outlet temperature
DNI
300 1200
DNI (W/m2)

200 800

100 400
#3 #4
0
0 10 20 30 40
Time (min)

Fig. 12. HTF flowrate and temperature variation curves in the field measurement.
(a) HTF temperature 320–570 K; (b) HTF temperature 490–620 K. Note: (1) curve of
flowrate was smoothed. (2) Images taken at time ‘#’ were present bellow.

images show there are no considerable deflections. The images #3,


#4 were taken in the afternoon while the HTF temperature was
high and the mass flowrate was very low, there are also no consid- Fig. 13. Images of the PTR during the field experiment.
erable deflections. By thoroughly checking all the images, the max-
imum deflection is found about 4 mm (see images #5 in Fig. 12),
which happened when the HTF flowrate is about 0.2 m/s and the
HTF temperature is about 630 K. The authors consider the deflec- 5. Conclusions and future work
tion is small and at the same scale of the measuring uncertainty.
Consequently, the authors can conclude there is no considerable In this study, an indoor experiment, numerical simulations (CFD
deflection of the stainless steel tube during the period of all the and FEA) and field measurements are conducted to jointly investi-
experiments. This agrees well with the FEA simulation results in gate the PTR’s structure reliability. The indoor experiment is con-
Section 4.2.2. ducted on a heat loss test-rig, and the deflection of the stainless
On this testing loop, other working conditions of the PTR, for steel tube under different isothermal conditions is studied. Numer-
example higher HTF temperature (to 630 K) and lower flowrate ical simulations are used to investigate the detailed temperature
(to 3.6 m3/h), have also been tested. The digital SLR camera did distribution, and its corresponding structural deformation of the
not detect any remarkable deflections of the stainless steel tube. PTR’s stainless steel tube. The simulations adapted a MCRT code,
The testing personnel also did not observe significant bending of FLUENT software and ANSYS workbench. Finally, field measure-
the stainless steel tube of all the 9 PTRs. Therefore, the authors ments are done on a 36 m long parabolic trough collector. The field
can conclude that the deflection (within the regime of elastic measurement monitors the stainless steel tube bending under dif-
deformation) from thermal stress is not the main cause of stainless ferent working conditions. Through this study, the following con-
steel tube bending. clusions can be drawn:
Of course, high temperatures will easily destroy the selective
coating, and possibly cause creep or plastic deformations of the (1) The HTF velocity and temperature have effects on the tem-
stainless steel tube. However temperatures sufficient to result in perature difference on a sectional plane of the stainless steel
this are rare and do not happen in practice. tube. The temperature difference changes inversely with
From the above results, and combined with the exist cases of HTF velocity and temperature.
PTR breakage and the field operational experiences, the authors (2) The HTF velocity and temperature have effects on the ther-
can now conclude with the elimination method that improper mal stress of the stainless steel tube of PTR. The thermal
installation and operational practices of parabolic trough receivers stress in the stainless steel tube changes inversely with
are the main causes of its structural failure. HTF velocity and temperature.
Z. Wu et al. / Applied Energy 123 (2014) 232–241 241

(3) When the PTR works under normal working conditions (HTF [15] Forristall R. Heat transfer analysis and modeling of a parabolic trough solar
receiver implemented in engineering equation solver. National Renewable
velocity 0.5–2 m/s, temperature 330–620 K), the measured
Energy Laboratory; 2003
stainless steel tube deflections are less than 4 mm. [16] Lei D et al. An experimental study of thermal characterization of parabolic
(4) The stainless steel tube deflection is weakly dependent on trough receivers. Energy Convers Manage 2013;69:107–15.
the thermal stress, and highly dependent on its supporting [17] Kalogirou SA. A detailed thermal model of a parabolic trough collector
receiver. Energy 2012;48(1):298–306.
conditions and the state of neighboring PTRs. [18] Li J et al. Vacuum reliability analysis of parabolic trough receiver. Sol Energy
Mater Sol Cells 2012;105:302–8.
Through experimental and numerical simulations, information [19] Li J et al. Hydrogen permeation model of parabolic trough receiver tube. Sol
Energy 2012;86(5):1187–96.
about PTR’s temperature, thermal stress of stainless steel tubes [20] Cheng ZD, He YL, Cui FQ. Numerical study of heat transfer enhancement by
and the stainless steel tube bending are obtained. This information unilateral longitudinal vortex generators inside parabolic trough solar
is of great importance to parabolic trough plant system design, the receivers. Int J Heat Mass Transfer 2012;55(21–22):5631–41.
[21] Wang P et al. Numerical study of heat transfer enhancement in the receiver
PTR’s structural reliability as well as installation and operation. In tube of direct steam generation with parabolic trough by inserting metal
the future, more field experiments regarding the installation and foams. Appl Energy 2012;102:449–60.
operation of PTRs will be arranged. In addition, field measurements [22] Muñoz J, Abánades A. Analysis of internal helically finned tubes for parabolic
trough design by CFD tools. Appl Energy 2011;88(11):4139–49.
will be conducted to investigate the PTR’s structural reliability. [23] Wang F et al. Thermal stress analysis of eccentric tube receiver using
concentrated solar radiation. Sol Energy 2010;84(10):1809–15.
[24] Shuai Y et al. Ray-thermal-structural coupled analysis of parabolic trough solar
Acknowledgements collector system. In: Solar collectors and panels, theory and
applications. Rijeka: Sciyo-Publishing Inc.; 2010. p. 341–56. p. 341–356.
[25] ANSYS F. Fluent 14.0 Documentation. Lebanon, NH: ANSYS Inc.; 2012.
This project was financially supported by the National Nature [26] Workbench A. 14.0. Computer software. Canonsburg, PA; ANSYS: 2012.
Science Foundation of China (Grant No. 51106148, No. 51006096 [27] He YL et al. A MCRT and FVM coupled simulation method for energy
and No. 51106150) and National Basic Research Program of China conversion process in parabolic trough solar collector. Renew Energy
2011;36(3):976–85.
(Grant No. 2010CB227106). [28] Wirz M, Roesle M, Steinfeld A. Three-dimensional optical and thermal
numerical model of solar tubular receivers in parabolic trough
concentrators. J Sol Energy Eng Trans ASME 2012;134(4).
References [29] Wu Z et al. Three-dimensional numerical study of heat transfer characteristics
of parabolic trough receiver. Appl Energy 2014;113:902–11.
[30] Cheng ZD et al. Three-dimensional numerical study of heat transfer
[1] Mills D. Advances in solar thermal electricity technology. Sol Energy
characteristics in the receiver tube of parabolic trough solar collector. Int
2004;76(1):19–31. Commun Heat Mass Transfer 2010;37(7):782–7.
[2] Sargent, Lundy. N.R.E. Laboratory, Executive summary. Assessment of
[31] Cheng ZD et al. Numerical simulation of a parabolic trough solar collector with
parabolic trough and power tower solar technology cost and performance nonuniform solar flux conditions by coupling FVM and MCRT method. Sol
forecasts. National Renewable Energy Laboratory; 2003. Energy 2012;86(6):1770–84.
[3] Kutscher C et al. Line-focus solar power plant cost reduction plan. Contract
[32] Schlichting H, Gersten K. Boundary-layer theory. Springer; 2004.
2010;303:275–3000. [33] Rohsenow WM, Hartnett JP, Cho YI. Handbook of heat transfer, vol. 3. New
[4] Lei D, Wang Z, Li J. The calculation and analysis of glass-to-metal sealing stress
York: McGraw-Hill; 1998.
in solar absorber tube. Renew Energy 2010;35(2):405–11. [34] Tannehill JC, Anderson DA, Pletcher RH. Computational fluid mechanics and
[5] Lei D, Wang Z, Li J. The analysis of residual stress in glass-to-metal seals for heat transfer. Taylor & Francis Group; 1997.
solar receiver tube. Mater Des 2010;31(4):1813–20. [35] Delgado-Torres AM, García-Rodríguez L. Comparison of solar technologies for
[6] Lei D et al. Experimental study of glass to metal seals for parabolic trough driving a desalination system by means of an organic Rankine cycle.
receivers. Renew Energy 2012;48:85–91.
Desalination 2007;216:276–91.
[7] Kennedy C, Price H. Progress in development of high-temperature solar- [36] Aoki K, Sone Y, Waniguchi Y. A rarefied gas flow induced by a temperature
selective coatings. Sol Eng 2005;2006:749–55. field: Numerical analysis of the flow between two coaxial elliptic cylinders
[8] Price H, Hale MJ, Mahoney R, et al. Developments in high temperature with different uniform temperatures. Comput Math Appl 1998;35(1):15–28.
parabolic trough receiver technology. [C]//ASME 2004 International Solar [37] Sharipov F, Bertoldo G. Heat transfer through a rarefied gas confined between
Energy Conference. American Society of Mechanical Engineers; 2004. p.
two coaxial cylinders with high radius ratio. J Vac Sci Technol A: Vac Surfaces
659–67. Films 2006;24(6):2087–93.
[9] Kennedy CE. Review of mid-to high-temperature solar selective absorber
[38] Roesle M et al. Analysis of conduction heat loss from a parabolic trough solar
materials. Golden, Colorado: National Renewable Energy Laboratory; 2002. vol. receiver with active vacuum by direct simulation monte carlo. Numer Heat
1617. Transfer Part A: Appl 2012;62(5):432–44.
[10] Zhang Q-C. Recent progress in high-temperature solar selective coatings. Sol [39] Roesle M, Coskun V, Steinfeld A. Numerical analysis of heat loss from a
Energy Mater Sol Cells 2000;62(1):63–74. parabolic trough absorber tube with active vacuum system. J Sol Energy Eng
[11] Ambrosini A, Lambert TN, Bencomo M, et al. Improved high temperature solar
Trans ASME 2011;133(3).
absorbers for use in concentrating solar power central receiver applications. [40] Tao W. Numerical Heat Transfer (In Chinese). second ed. Xi’an: Xi’an Jiaotong
[C]//ASME 2011 5th International Conference on Energy University Press; 2001.
Sustainability. American Society of Mechanical Engineers; 2011. p. 587–94. [41] Churchill SW, Chu HHS. Correlating equations for laminar and turbulent free
[12] Kaluža L et al. Sol–gel derived CuCoMnOx spinel coatings for solar absorbers: convection from a horizontal cylinder. Int J Heat Mass Transfer
Structural and optical properties. Sol Energy Mater Sol Cells
1975;18(9):1049–53.
2001;70(2):187–201. [42] Hirsch T, et al. Stratified Flow Phenomena in Parabolic Trough Systems. In:
[13] Price H, Forristall R, Wendelin T, et al. Field survey of parabolic trough receiver
Solar Paces Symposium. Marrakech, Morocco; 2012.
thermal performance. [C]//ASME 2006 International Solar Energy [43] Schiricke B et al. Experimental verification of optical modeling of parabolic
Conference. American Society of Mechanical Engineers; 2006. p. 109–16. trough collectors by flux measurement. J Sol Energy Eng 2009;131:011004.
[14] Lüpfert E et al. Experimental analysis of overall thermal properties of parabolic [44] Bader R, Pedretti A, Steinfeld A. J Sol Energy Eng Trans ASME 2012;134(2).
trough receivers. J Sol Energy Eng, Trans ASME 2008;130(2):0210071–75.

Vous aimerez peut-être aussi