Vous êtes sur la page 1sur 10

Renewable Energy 36 (2011) 976e985

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

A MCRT and FVM coupled simulation method for energy conversion process
in parabolic trough solar collector
Ya-Ling He*, Jie Xiao, Ze-Dong Cheng, Yu-Bing Tao
State Key Laboratory of Multiphase Flow in Power Engineering, School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, PR China

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

Article history: A coupled simulation method based on Monte Carlo Ray Trace (MCRT) and Finite Volume Method (FVM)
Received 14 September 2009 is established to solve the complex coupled heat transfer problem of radiation, heat conduction and
Accepted 25 July 2010 convection in parabolic trough solar collector system. A coupled grid checking method is established to
Available online 17 September 2010
guarantee the consistency between the two methods and the validations to the coupled simulation
model were performed. Firstly, the heat flux distribution on the collector tube surface was investigated to
validate the MCRT method. The heat flux distribution curve could be divided into 4 parts: shadow effect
Monte Carlo Ray Trace
area, heat flux increasing area, heat flux reducing area and direct radiation area. The heat flux distri-
Finite Volume Method
Coupling heat transfer
bution on the outer surface of absorber tube was heterogeneous in circle direction but uniform in axial
Parabolic trough collectors direction. Then, the heat transfer and fluid flow performance in the LS-2 Solar Collector tube was
investigated to validate the coupled simulation model. The outlet temperatures of the absorber tube
predicted by the coupled simulation model were compared with the experimental data. The absolute
errors are in the range of 1.5e3.7  C, and the average relative error is less than 2%, which demonstrates
the reliability of the coupled method established in this paper. At last, the concentrating characteristics of
the parabolic trough collectors (PTCs) were analyzed by the coupled method, the effects of different
geometric concentration ratios (GCs) and different rim angles were examined. The results show the two
variables affect the heat flux distribution. With GC increasing, the heat flux distributions become gentler,
the angle span of reducing area become larger and the shadow effect of absorber tube become weaker.
And with the rim angle rising, the maximum value of heat flux become lower, and the curve moves
towards the direction 4 ¼ 90 . But the temperature rising only augments with GC increasing and the
effect of rim angle on heat transfer process could be neglected, when it is larger than 15 . If the rim angle
is small, such as qrim ¼ 15 , lots of rays are reflected by glass cover, and the temperature rising is much
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction From 1970s, some researchers have tried to use numerical

method to investigate the heat flux distribution characteristics of
The photo-thermal conversion process of parabolic trough solar the parabolic solar concentrators. Evens [2] and Harris et al. [3] both
thermal power generation system can be divided into the following developed integral relationships for evaluating the intensity
sections: solar radiation is collected by the parabolic trough distribution on the flat absorbers used with cylindrical parabolic
collectors (PTCs), then is reflected and focused to the outer surface solar concentrators. Considering the influence of the finite size of
of the absorber tube where the radiant energy is converted into the sun, the “cone optics” approach was used [2]. Then, Jeter [4,5]
thermal energy. Thermal energy will be conducted to the inner established a first integral of the concentrated radiant flux
surface of the absorber tube and be transferred by the heat transfer density for trough concentrators. Those works had given out
fluid inside the absorber tube with forced convection heat transfer concentrating characteristics of different parabolic solar concen-
[1]. This is a coupled heat transfer problem with complex geometry trators. But all those integral approaches needed special integration
condition. routine, meanwhile, the optical parameters and geometrical
properties of collectors also couldn’t be changed conveniently.
Monte Carlo Ray Trace (MCRT) method is a very flexible method,
which can be used to solve this coupled problem. The principles of
* Corresponding author. Tel.: þ86 29 8266 3300; fax: þ86 29 8266 7745. the MCRT method applied to the radiative heat transfer can be
E-mail address: yalinghe@mail.xjtu.edu.cn (Y.-L. He). found from textbooks [6]. This method was used for simulating the

0960-1481/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved.
Y.-L. He et al. / Renewable Energy 36 (2011) 976e985 977

reflection process of solar radiation [7], dish solar concentrator [8]

and solar chemical reactor [9], etc.
There are also many published studies about the heat transfer
process between the absorber tube wall and working fluid. A
detailed thermal model was established to calculate the heat loss
of trough collector [10], in which the solar radiation was regarded as
a constant and the influence of nonlinearity on heat loss was
also ignored. Nowadays, this model is also used to analyse the
thermal efficiency of trough system [11]. In order to analyse the
two-phase flow region in direct steam generation (DSG), some 2D
simulation methods were also established to study the temperature
distribution along the cross-section of absorber tube. In those
methods, the nonlinearity of heat flux was considered, but simpli-
fied distribution assumptions were used to simplify the heat flux
distribution around the tube wall. For example, Martinez et al. used
a sectional heat fluxes as boundary conditions of Finite-Difference
Method (FDM) to solve energy equations [12]. Eck et al. used the
Finite-Element Method (FEM) program ANSYS with a typical flux
Fig. 1. Schematic of a parabolic trough collector.
distribution boundary to calculate the temperature profile for the
LS-3 type collector [13], then they chose a Gaussian distribution
assumption as the flux distribution boundary for Finite-Element
2.2. Details of MCRT
Method (FEM) analysis and a rectangular distribution assumption
as the flux distribution boundary for a simplified analytical
The flowchart for MCRT is shown in Fig. 3. The diamond boxes
solution [14].
can be considered as judgments for position relation of straight and
Up to now, few papers have focused on the flow and heat
curve, such as judgments of “Shadowed region by absorber”,
transfer characteristics of working fluid inside the absorber tube
“Reflected region by parabola”, “Hit glass tube” or “Hit absorber
under the realistic non-uniform heat flux distribution boundary
tube”. If those judgments are “YES”, the hitting locations are
conditions. In this paper, the realistic non-uniform heat flux
determined, such as processes of “Count hitting position on glass
distribution is used as the boundary condition to simulate the heat
tube and absorber tube”. All those judgments and processes are
transfer process inside the absorber tube, and the characteristics
easy events as solving quadratic equations. So details of those boxes
analysis under the influence of the non-uniform heat flux distri-
are not shown in following sections. However, in several processes,
bution was made. For simulation, a coupled numerical method of
new trace directions should be decided, such as “Initialize photon
MCRT and FVM was established to obtain the 3D flow fields and
distribution”, “Reflect” and “Transmit”. Monte Carlo Method is used
temperature distribution for the coupled heat transfer problem in
in those sections which will be detailed in following sections. Other
absorber tube with the complexity of non-uniform heat flux
processes which are statistic events, such as “Absorb photon”,
boundary condition. In order to validate the coupled simulation
“Count photon distribution” and “Count heat flux distribution”, are
method, the concentrating characteristics and heat transfer in
also discussed in following sections.
absorber tube of the LS-2 Solar Collector were simulated and
compared with test data [15]. Then the model was used to
2.2.1. Initialize photon distribution
simulate the influence of the nonlinear heat flux distribution on
Initializing photon distribution is used to decide the location of
overall heat transfer performance. The effects of different
photon hitting the parabola, the direction of photon tracing and the
geometric concentration ratios and rim angles are also discussed
energy weight carried by each photon packet. The location is
in this paper.
described by the Cartesian coordinates (x,y,z) and the direction is
described by the directional cosines (ux,uy,uz) [16]. The location
2. Simulation method
where photon hits the parabola is specified as follows:

2.1. The physical model and coordinate systems

Fig. 1 is a schematic of traditional PTC, which is made by bending

a sheet of reflective material into a parabolic shape surface. A
receiver tube is placed along the focal line of the parabolic reflective
collector. The receiver tube usually includes an inner absorber tube
and a glass cover. The glass cover is used for reducing heat loss; as
well as the annular space between the inner tube and the glass
cover is vacuumed, which helps to decrease the convection heat
loss [1].
Two coordinate systems are used in this paper. The Cartesian
coordinate system is used for tracing photon movements in
MCRT code and solving correlated equations in FLUENT, as shown
in Fig. 1. In the cross-section of absorber tube, a cylindrical
coordinate system is used to count the photon distribution and
heat flux distribution (Fig. 2). Because the absorber tube is
located along the focal line and the parabola is symmetric, the
photon and heat flux distribution is symmetric along the z axis.
So the range of circle angle 4 is from 90 to 90 . Fig. 2. The cylindrical coordinate system.
978 Y.-L. He et al. / Renewable Energy 36 (2011) 976e985

Initialize photon The energy weight, w, is also initialized, which carried by each
distribution photon packet. It is a non-dimensional energy parameter and set
the value as w ¼ 1.
Shadowed by absorber?
Here, each photon packet is initially assigned a weight, w. In this
paper, the weight is set as w ¼ 1.
2.2.2. Reflection
Reflected by parabola? Two steps are established to complete the reflection process.
Y First, judge whether the photon is reflected. Second, determine the
Reflect new photon trace direction after reflection.
In first step, another random number, x5, is generated; and then
N the random number is compared with the reflectivity of parabolic
Hit glass tube?
collector, Rc. If x5  Rc, the photon is reflected by parabolic trough; if
Y not, the photon will not survive.
N The theory of reflect was shown in Ref. [7]. A specular reflection
Transmit? hypothesis is used in this paper. The directional cosines are speci-
Y fied as follows:
Count hitting position 8 0
on glass tube >
> u ¼ ux " #," !#
> x
> 0 0 y0 y0 2
< uy ¼ uy  y sinðqÞsinð4Þ þ cosðqÞ f 1þ 2
Count new moving 2f 4f
direction (5)
> " #, !
> 0 y0 2
> u0 ¼ uz þ 2 y sinðqÞsinð4Þ þ cosðqÞ
N >
: z 1 þ
Hit absorber tube? 2f 4f 2

Count hitting position 2.2.3. Transmission

on absorber tube After the photon hitting the glass tube, whether the photon
transmits across the glass tube should be decided. The glass
Absorb photon property and the photon tracing direction both affect the trans-
mission. The effect of glass property on transmission is the same as
N it to the reflection. A random number, x6, is used to compare with
Last photon?
the transmittance of glass, Tg. If x6  Tg, the photon transmits across
Y the glass. While, according to Snell’s law, the probability of a photon
Count photon transmission is related to the angle of incidence, ai and the
refractive indices of the media that the photon is incident from, ni
and transmitted to, nt. Details are shown in Ref. [16].
Count heat flux
distribution Because the thickness of glass is ignored, the position of the
photon leaving the glass tube could be considered to be the same
position where it hits the glass, and the direction cosines also do
not change.

Fig. 3. Flowchart for MCRT.

2.2.4. Photon statistics
8 An array p[ie] is used to count the distribution of photon on the
< x0 ¼ x0 þ x1 $xL
outer surface of the absorber tube. Where ie is in the range:
y0 ¼ y0 þ x2 $yL (1)
: 0 1  ie  Ne, and Ne is the number of elements. The grid used in this
z ¼ z0 þ y2 =ð4f Þ
paper is generated by Gambit software. The mesh information, such
In this paper, x is used as random numbers generated by as the coordinate values of each node and the ordinal of nodes,
computer, which is uniformly distributed between the intervals could be obtained, by which the elements could be defined. If the
(0,1). Position (x0,y0,z0) is initialized to (0,(yL/2),f). A deflection photon hit the absorber tube at point p, the position can be
angle q (0 q <p) and an azimuth angle f (0  f < 2p) are used to determined by an array m(x00 ,y00,z00 ). Then judge which element the
decide the directional cosines: photon deposits in by counting areas. For example, one element
8 composed by 4 nodes, each node could marked as A, B, C and D.
< ux ¼ sinðqÞcosðfÞ Connection the conjoint 2 points and the point p can form a triangl..
u ¼ sinðqÞsinðfÞ (2)
: y
uz ¼ cosðqÞ Table 1
Parameters of SEGS LS-2 PTCs [15].
The azimuth angle is specified as follows:
Parameter Value
f ¼ 2px3 (3) Parabolic trough reflector xL  yL 7.8 m  5 m
f 1.84 m
Because the sun radiations are not parallel exactly, deflection Rc 0.93
angle q should be chosen in an optic cone. Here, qsun ¼ 16 is used to Receiver rg 0.0575 m
represent the finite size of the sun. So the angle q is specified as: ra 0.0350 m
Tg 0.95
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi aa 0.96
q ¼ arcsin sin q2sun $x4 (4) e 0.002 m
Y.-L. He et al. / Renewable Energy 36 (2011) 976e985 979

Fig. 4. Schematic of systematic error in the grid system.

Then plus the triangle areas of ABP, BCP, CDP and DAP together, and v v mt vk
marked as S0i . Meanwhile, the area of each element could be count ðrui kÞ ¼ mþ þ Gk  r3 (8)
vxi vxi sk vxi
out as Si. The value of S0i and Si are compared. If jS0i  Si j  s, it could
be considered that the photon deposits in this element. s is an
infinitesimal, which has same dimension as Si. In this paper, the   
value of s is set as s ¼ 5  107. Then the value of p[ie] is renewed as:
v v mt v3 3
ðrui 3Þ ¼ mþ þ ðc1 Gk  c2 r3Þ (9)
vxi vxi s3 vxi k
p½ie )p½ie  þ w$aa (6)
The turbulent viscosity mt is specified:

2.2.5. Heat flux distribution mt ¼ Cm r (10)
The photon distribution, p[ie], is dimensionless. It only presents 3
the distribution characteristic of photon, but not the exact value of
heat flux. The heat flux distribution is specified as q[ie]: Gk denotes the production rate of k, which is specified as:

qsun $xL yL vu vui vuj
q½ie  ¼ p½ie $ (7) Gk ¼ mt i þ (11)
Nr Si vxj vxj vxi

2.3. Coupled method of MCRT and FVM The governing equations for continuity, momentum and energy
can be expressed as follows:
2.3.1. Governing equations for FVM
The fluid flow is turbulent and in steady state at test conditions Continuity equation:
shown in Table 1. The governing equations include the continuity,
momentum, energy equations and the standard ke3 two-equation v
ðrui Þ ¼ 0 (12)
model. vxi
The standard ke3 two-equation turbulence model can be
expressed as follows: Momentum equation:

" ! #
v  vp v vui vuj 2 vul
rui uj ¼  þ ð tþ Þ
m m þ  ð tþ Þ
m m d
60 vxi vxi vxj vxj vxi 3 vxl ij
Local Concentration Ratio

1 2 3 4
50 þ rgi ð13Þ
Jeter's result
This paper
30 Table 2
Typical test data [15].
Test conditions qsun (W m2) Qm (kg s1) Tin ( C) Tout ( C)

10 Case 1 933.7 0.6782 102.2 124.0

Case 2 937.9 0.6206 297.8 316.9
Case 3 920.9 0.5457 379.5 398.0
0 Case 4 880.6 0.6205 299.0 317.2
-90 -75 -60 -45 -30 -15 0 15 30 45 60 75 90
Case 5 909.5 0.6580 250.7 269.4
Case 6 968.2 0.6536 151.0 173.3
Case 7 982.3 0.6350 197.5 219.5
Fig. 5. Local concentration ratio distribution.
980 Y.-L. He et al. / Renewable Energy 36 (2011) 976e985

Table 3
f and Nu for different grid number systems.

Grids number(Nc  Nx) f Nu

68  320 0.01064856 91.03302
68  400 0.01064735 91.07156
68  480 0.01064366 91.11041
80  320 0.01063134 91.04913

Energy equation:

v v m mt vT
ðrui TÞ ¼ þ þ qR (14)
vxi vxi Pr sT vxi

The standard constants are employed, Cm ¼ 0.09, c1 ¼1.44,

c2 ¼ 1.92 sk ¼ 1.0, s3 ¼ 1.3, sT ¼ 0.85[17].
Since the ke3 turbulence model is not applicable in regions close Fig. 6. Heat flux distribution on outer surface of absorber tube.
to solid walls where viscous effects predominate over turbulent
ones, the wall functions method is used [17].
In this equation, DSi is related to the geometric parameters r, xL
2.3.2. Transferring the MCRT information to FLUENT and grid parameters Nc, Nx. For a specific absorber tube, the
In order to transfer the information between MCRT and Fluent geometric parameters would not change. So in equation (15), the
conveniently, the same meshes are needed. The discrete non- DSi is a function of the grid parameters (Nc,Nx). The value of DSi
uniform heat flux distribution calculated by MCRT is considered to should be sufficiently small to ensure the photon would not be
be the second type boundary condition for this problem. A profile of double counted on the surface of the absorber tube. Theoretically,
total surface heat flux is exported from MCRT, and the heat flux the value of DSi is the smaller, the better. But actually, during the
distribution is added to the profile. Then, the new profile is used to calculation process, the DSi should set an appropriate constant
be the boundary condition of the outer surface of the absorber tube. value. In our MCRT code, variables are single precision, and
In reality, the solar absorption in the absorber (opaque metal accounting the impact of rounding errors, the value of DSi is chosen
material) is volumetric phenomena [18]. However, most of the less than 5  107. Then a combination of Nc and Nx was chosen as
absorption in the absorber occurs very close to the surface (about the minimum grid system to reach the requirement of systematic
6A)[18]. Therefore, the error in treating solar absorption as error.
a surface phenomenon should be relatively small [19].
2.4.2. Grid independence test
2.4. A coupled grid checking method The average friction coefficient f and the average Nusselt
Number Nu are used here for grid independence test. It is a tradi-
2.4.1. The minimum grid density for MCRT tional grid independence test method for the fluid and heat transfer
In the MCRT code, attention should be paid to the systematic problem. It should be noted that the minimum grid density is
error in certain areas, in this area, the S0i is always greater than Si. limited by the systematic error discussed above.
This is because in the grid system generated by Gambit software,
the circles are substituted by the approximate polygons. According 2.4.3. Suitable photon packets number
to Section 2.2.4, the hitting position p is on the cylindrical surface, After the grid independence test, a suitable grid system is
but there is a small distance between the point and the element chosen. Then a suitable photon packets number should be chosen
ABCD (Fig. 4). The small distance causes the numerical difference of on this grid system to meet the accuracy requirement of MCRT
S0i and Si, which is the systematic error. Here, DSi is defined as:
DSi ¼ S0i  Si . If p is in the middle of cylindrical surface ABCD, DSi
can get maximal value: 4
  rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4
6x10 1 2 3 4
p p 2 x2L rxL p
DSi ¼ 2r sin r 1  cos
2 þ 2þ 2  2 cos
Nc Nc 4Nx Nx Nc 5x10

xL p
q / W m-2

 2r sin ð15Þ 4x10


Nx Nc

Table 4 2x10
Photon packets checking in MCRT.
Number of photon packets dq 1x10
1.0  106 0.5134497 0
5.0  106 0.2395571 -90 -75 -60 -45 -30 -15 0 15 30 45 60 75 90
1.0  107 0.1627726
2.5  107 0.0939214
5.0  107 0.0538396
Fig. 7. Heat flux distribution on circle direction.
Y.-L. He et al. / Renewable Energy 36 (2011) 976e985 981

Table 5 been tested in details in Ref. [15], were analyzed by the coupled
Comparison of the outlet temperature between simulation and test result. method, to verify the coupled method is correct.
Tout ( C) 0 ð CÞ
Tout 0
Tout  Tout ð CÞ d
Case 1 124.0 126.2 2.2 0.01847 3.1. MCRT code checking
Case 2 316.9 318.4 1.5 0.00499
Case 3 398.0 400.4 2.4 0.00563 In order to confirm the accuracy of our MCRT code, numerical
Case 4 317.2 319.9 2.7 0.00851
results were compared with Jeter’s [5]. In Jeter’s paper, a parameter
Case 5 269.4 272.4 3.0 0.00811
Case 6 173.3 177.0 3.7 0.02135 called Local Concentration Ratio (LCR) was discussed. LCR is
Case 7 219.5 222.8 3.3 0.01503 a relative parameter that is similar to the heat flux distribution in
this paper. The relationship of the two parameters is shown as [5]:

method. A group of heat flux distributions, q[ie], are calculated by q ¼ LCR$qsun (18)
using different numbers of photon packets, then the maximum The geometric shape of PTC and the parameters used here are
photon packets is chosen to be the true value, q½i _ e . The next step is the same as used in Jeter’s paper. The rim angle qrim and the
to analyze the errors of q[ie] in each nodes, marking as dq[ie]. At last, geometric concentration ratios GC are specified as follows:
the average heat flux errors, dq , is calculated. If dq < s2 , the number
of photon packets is considered to be the suitable one. dq[ie] and dq YL
GC ¼ (19)
are specified as follows: 2pra

_ e j
jq½ie   q½i  
dq ½ie  ¼ (16) 1
_ e
q½i qrim ¼ arctan (20)
ð2f =YL Þ  ðYL =8f Þ
PNe The result is shown in Fig. 5, and the two curves follow the same
d ½i 
ie ¼ 1 q e trend. The maximum and minimum values are also very close,
dq ¼ (17)
Ne which verify that the present code is reliable.
As shown in Fig. 5, the curve could be divided into four parts,
3. Simulation method verification marked by 1, 2, 3 and 4. In part 1, because of the sun’s radiation is
shadowed by the receiver; the heat flux is much lower but increases
In order to verify the accuracy of this coupled simulation rapidly, which is defined as the shadow effect area. In part 2, more
method, the simulation results were compared with references. and more rays are reflected to the receiver, so the heat flux
Firstly, the MCRT code was checked by comparing our results to increases steady, which is defined as the heat flux increasing area.
Jeter’s [5]. And then, the concentrating characteristics and heat In part 3, the heat flux reduces rapidly with the rays reflected
transfer in absorber tube of the LS-2 Solar Collector, which have decrease because of the sun shape qsun, which is defined as the heat

Fig. 8. 3D heat flux distribution under different GCs.

982 Y.-L. He et al. / Renewable Energy 36 (2011) 976e985

flux reducing area. At last, in part 4, there are few reflected rays In Dudley et al.’s report, many measured efficiency test data
reaching the receiver, the absorber tube only receives the direct were shown for different cases. Seven typical cases were chosen
radiation of sun, the heat flux distribution is in a very low level, here for simulation and validation, which are shown in Table 2.
which is defined as the direct radiation area. Here, a new variable 40
is defined as 40 ¼ qrim  90 , which is in the center of part three. 3.2.2. Grid checking
According to Section 2.4.1, taking Case 1 in Table 2 as example
for grid checking, the minimum grid density for the absorber tube
3.2. Coupled method validation by simulation for LS-2 Solar
was calculated to be: Nc  68 and Nx  320.
After that, the grid independence test was further made in
FLUENT. Four different grid systems were investigated. They were:
In this section, this coupled simulation method is applied to
68  320, 68  400, 68  480 and 80  320. The average friction
simulate the concentrating characteristics and heat transfer in
coefficient and the average Nusselt number of different grid
absorber tube of the LS-2 Solar Collector, which have been tested in
systems are shown in Table 3. From this table, it could be seen that
Ref. [15].
there are no significant differences between those grid systems. So
the grid system 68(Nc)  320(Nx) could be regarded as grid-
3.2.1. Physical model and parameters
The LS-2 parabolic trough collector module tested at Sandia
At last, the photon packet checking was made. 1 108 photon
National Lab was chosen for simulation, parameters are shown in
packets were chosen as the true value. And then the photon packets
Table 1 [15]. The computational domain includes the absorber tube
of 1.0  106, 5.0  106, 1.0  107, 2.5  107 and 5.0  107 were
domain (solid), the liquid oil domain (fluid) and the flow restriction
investigated. The average heat flux errors are shown in Table 4, and
device domain (solid). The heat conduction in the tube and the flow
5.0  107 was considered to be the suitable value.
restriction device are taken into account, in which the temperature
distribution will be determined by the full-field solving method.
3.2.3. Heat flux distribution
The material for the solid domain is stainless steel and the thermal
The heat flux distribution on the outer surface of absorber tube
conductivity is fixed at 54 W m1 K1.
is shown in Figs. 6 and 7. The asymmetric distribution of heat flux in
Sylitherm 800 liquid oil was used as the working fluid. And the
circle direction is obviously, but the heat flux distribution in axial
properties of working fluid, such as: isobaric specific heat capacity
direction (x) is uniform, as shown in Fig. 6. Furthermore, the curve
(cp), thermal conductivity (l), density (r) and dynamic viscosity (m)
in Fig. 7 can also divided into four parts as shown in Section 3.1.
should be selected to simulate the heat transfer process in absorber
tube by FLUENT. Those properties are temperature-dependent and
3.2.4. Outlet temperature
derived from the following correlations for temperatures range
Seven cases as shown in Table 1 were simulated using the
from 373.15 to 673.15 K [20].
couple method [15]. The predicted outlet temperatures are shown

in Table 5. Comparing with the results in Dudley et al.’s report, the
cp ¼ 0:001708T þ 1:107798 kJ kg1 K (21) absolute errors are in the range of 1.5e3.7  C, the maximum rela-
tive error is 0.02135, and the average relative error is 0.01325 (less
l ¼ 5:753496  1010 T 2  1:875266  104 T þ 1:900210 than 2%). It is a good agreement of simulation result and test result,

 101 W m1 K (22) which proves that the models and methods used in the present
study are feasible and the numerical results are reliable.

It also can be seen from Table 5 that the outlet temperature Tout
r ¼ 4:153495  101 T þ 1:105702  103 kg m3 (23) obtained by the numerical simulation is larger than the corre-
sponding test result Tout. The error is mainly due to the assumption
of neglecting the convective heat losses between the outer wall of
m ¼ 6:672  107 T 4  1:566  103 T 3 þ 1:388$T 2  5:541 absorber tube and the inner surface of glass cover tube, and the
 102 T þ 8:487  104 ðmPa sÞ ð24Þ assumption of total absorption of the solar energy arrive at the
outer surface of the inner absorber tube in the MCRT code and so
on. On the other hand, there also could be some measurement
errors constitutionally.
GC =10
GC =30
GC =50 25.0
q / W m-2



0.0 23.0
-90 -75 -60 -45 -30 -15 0 15 30 45 60 75 90 10 30 50
Geometric Concentration ratio
Fig. 9. Heat flux distribution under different GCs on circle direction. Fig. 10. The temperature rising under different GCs.
Y.-L. He et al. / Renewable Energy 36 (2011) 976e985 983

Fig. 11. 3D heat flux distribution of different rim angles.

4. Characteristics analysis for PTCs

The coupled simulation method can be used in the concen- o =15
=-60 rim
trating characteristics analysis for PTCs. In this section, the varia- 5
1.0x10 o
tions of heat flux distributions in different GC and different rim
=-45 =30
angle are discussed. Meanwhile, in those conditions, the tempera- 4
8.0x10 =-30
o =45
ture rising of fluid will also be shown. For investigating the influ- o
q / W m-2

o =60
ences of GC and rim angle, the total sun radiation is kept constant, 4 =-15 rim
6.0x10 =75
which means the aperture yL would not change and set as yL ¼ 5 m. rim
For the same reason, the mass flow rate was set to a constant as 4 =90
4.0x10 rim
0.6872 kg s1, which was a test value in Dudley et al.’s report [15]. =0
4.1. Influence of different geometric concentration ratios
For holding the aperture yL constant, according to equation (19), -90 -75 -60 -45 -30 -15 0 15 30 45 60 75 90
the absorber tube radius ra should be changed. For the GCs at 10, 20, o
30, the radiuses of absorber tubes are 0.080 m, 0.027 m, 0.016 m
respectively, with yL ¼ 5 m. Here, the rim angle is chosen as 90 . Fig. 12. Heat flux distribution under different rim angles on circle direction.
984 Y.-L. He et al. / Renewable Energy 36 (2011) 976e985

24.0 To verify the MCRT code, the simulation results were compared
with reference and the two curves matched well. The results show
23.5 23.57K 23.36K 23.35K that the curve of heat flux distribution is divided into 4 parts:
shadow effect area, heat flux increasing area, heat flux reducing
23.44K 23.33K
23.0 area and direct radiation area. The heat flux distribution on the
outer surface of absorber tube was heterogeneous in circle

22.5 direction but uniform in axial direction (x). Then, the coupled
method was used to simulate the LS-2 Solar Collector for method
verification. And the outlet temperatures of four cases were
counted out and compared with Dudley et al.’s report data. The
comparison results show they are in good agreement, which
validate the coupled method established in this paper is useful
and credible.
0 15 30 45 60 75 90 The concentrating characteristics of PTCs were analyzed using
o the coupled method. The variations of heat flux distributions in
/ different GCs and different rim angles were discussed. With GC
increasing, the heat flux distributions become gentler, the angle
Fig. 13. The temperature rising under different rim angles.
span of reducing area become larger and the shadow effect of
absorber tube become weaker. The temperature rising also
improved with GC increasing. And with the rim angle rising, the
Figs. 8 and 9 show the heat flux distributions on the outer surface of
maximum value of heat flux becomes lower, and the curve moves
the absorber tubes under different GCs. As shown in those figures,
towards the direction 4 ¼ 90 . But there is small effect on heat
with GC rising, the heat flux distributions become gentler, the angle
transfer process. If the rim angle is small, such as qrim ¼ 15 , lots of
span of reducing area become larger and the shadow effect of
rays are reflected by glass cover, and the temperature rising is much
absorber tube become weaker.
The HTF temperature rising under different GCs is shown in
Fig. 10. Because of the total radiation and mass flow rate both set to
constant, the variation of temperature rising is only related to the
different heat flux distributions, which is caused by different GCs. As
The present work is supported by the National Natural Science
shown in Fig. 10, with GC rising, the temperature rising improved.
Foundation of China (Nos. 50736005, 50906070) and National Basic
Gentler heat flux distribution made the heat lost reduced.
Research Program of China (973 Program) (2010CB227100).

4.2. Influence of different rim angles Nomenclature

In this section, the aperture is also kept constant, and the

absorber tube radius is set at ra ¼ 0.035 m. Figs. 11 and 12 showed cp isobaric specific heat capacity (kJ/kg K)
the heat flux distribution on the outer surface of the absorber tube Cm, c1, c2 Coefficients in turbulence model
under different rim angles. With qrim rising, the maximum value of e Thickness of the absorber tube (m)
heat flux become lower, and the curve moves towards the direction f Focal length (m)
of 4 ¼ 90 , except qrim ¼ 15 . The center of reducing area, 40, is f Average friction coefficient
a function of qrim. With qrim rising, 40 rises, and then, the concen- GC Geometric concentration ratios
tration area becomes larger, which causes the maximum value of k Thermal conductivity or kinetic energy or von Karman
heat flux reduce. It is noteworthy that the heat flux distribution of constant
qrim ¼ 15 does not equate to the distribution rule of others. It is kw Thermal conductivity of the inner absorber tube wall
much lower compared to others. A considered reason is that the rim (W/m K)
angle is so small, then the angles between reflect rays and normal kin Inlet turbulent kinetic energy
line of absorber tube is large, which causes lots of rays are reflected ni Refractive indices
by the glass cover, and less rays absorbed by the tube. Ne Number of elements
Fig. 13 shows the temperature rising under different rim angles. Nn Number of nodes
There is no significant difference of the temperature rising which rim Nr Number of ray
angle are qrim ¼ 30 , 45 , 60 , 75 , 90 . It seems that the influence of Nc Number of nodes in circumferential direction
rim angle to heat transfer process is also small. An exception is Nx Number of nodes in x direction
qrim ¼ 15 , the temperature rising is much smaller than other Nu Average Nusselt number
conditions, that is because less rays traced arriving at absorber tube. p[ie] Discrete photon flux distribution
q Heat flux distribution
5. Conclusion qsun Direct normal insolation
q[ie] Discrete heat flux distribution
In this paper, a coupled method of MCRT and FVM is established _ e
q½i True value of heat flux distribution
to simulate the photo-thermal process of parabolic trough solar Qm Mean mass flow rate (kg/s)
thermal power generation. The MCRT code is used to gain the r Radius
heterogeneous heat flux distribution on absorber tube, and FVM ra Radius of absorber tube
software FLUENT is used to solve the fluid and heat transfer rg Radius of glass cover
problem. For the same mesh systems used in MCRT and FLUENT is R The reflectivity of reflector
convenient for the information transfer, a coupled grid checking Si Area of element (m2)
method is also established. Tg Transmittance of glass
Y.-L. He et al. / Renewable Energy 36 (2011) 976e985 985

ux,uy,uz Directional cosines [2] Evens DL. On the performance of cylindrical parabolic solar concentrators
with flat absorbers. Solar Energy 1977;19:379e85.
w Weight
[3] Harris James A, Duff William S. Focal plane flux distribution produced by solar
x,y,z Cartesian coordinate concentrating reflectors. Solar Energy 1981;27:403e11.
xL Length of parabolic in x direction [4] Jeter MS. Calculation of the concentrated flux density distribution in para-
yL Length of parabolic in y direction bolic trough collectors by a semifinite formulation. Solar Energy 1986;
[5] Jeter MS. Analytical determination of the optical performance of practical
parabolic trough collectors from design data. Solar Energy 1987;39:11e21.
Greek symbols [6] Modest MF. Radiative heat transfer. 2nd ed. New York: Academic Press; 2003.
ai Incidence angle [7] Riveros HG, Olova AI. Graphical analysis of sun concentrating collectors. Solar
at transmission angle Energy 1986;36:313e22.
[8] Shuai Y, Xia XL, Tan HP. Radiation performance of dish solar concentrator/
aa absorptance
cavity receiver systems. Solar Energy 2008;82:13e21.
d error [9] Kraupl S, Steinfeld A. Monte Carlo radiative transfer modeling of a solar
dq[ie] error of heat flux distribution chemical reactor for the co-production of zinc and syngas. Journal of Solar
dq average heat flux error Energy Engineering 2005;127:102e8.
[10] Odeh SD, Morrison GL, Behnia M. Modelling of parabolic trough direct steam
3 turbulent dissipation rate or emissivity generation solar collectors. Solar Energy 1998;62:395e406.
3in inlet turbulent dissipation rate [11] Hou Z, Zheng D, Jin H, Sui J. Performance analysis of non-isothermal solar
3w emissivity of the inner tube wall reactors for methanol decomposition. Solar Energy 2007;81:415e23.
[12] Martinez I, Almanza R. Experimental and theoretical analysis of annular two-
f Azimuthal angle phase flow regimen in direct steam generation for a low-power system. Solar
4 circle angle Energy 2007;81:216e26.
l thermal conductivity (W/m K) [13] Eck M, Steinmann WD, Rheinl J. Maximum temperature difference in hori-
zontal and tilted absorber pipes with direct steam generation. Energy 2004;
m dynamic viscosity (Pa s) 29:665e76.
mt turbulent viscosity (Pa s) [14] Eck M, Steinmann W. Modelling and design of direct solar steam generating
q deflection angle collector fields. Journal of Solar Energy Engineering 2005;127:371e80.
[15] Dudley VE, Kolb GJ, Sloan M, Kearney D. Segs Ls2 solar collector-test results.
qsun finite size of the sun USA: SANDIA; 1994.
qrim the rim angle [16] Wang L, Jacques SL, Zheng L. McmleMonte Carlo modeling of light transport
r density (kg/m3) in multi-layered tissues. Computer Methods and Programs in Biomedicine
sT turbulent Prandtl number
[17] Launder BE, Spalding DB. The numerical computation of turbulent flows.
sk,s3 turbulent Prandtl numbers for diffusion of k and e Computer Methods in Applied Mechanics and Engineering 1974;3:269e89.
x Random number [18] Incropera F, Dewitt D. Fundamentals of heat and mass transfer. 3rd ed. New
York: John Wiley and Sons; 1990.
[19] Forristall R. Heat Transfer Analysis and Modeling of a Parabolic Trough Solar
References Receiver Implemented in Engineering Equation Solver. Report, NREL; 2003.
[20] Delgado-Torres AM, Garc RL. Comparison of solar technologies for driving
[1] Kalogirou SA. Solar thermal collectors and applications. Progress in Energy a desalination system by means of an organic rankine cycle. Desalination
and Combustion Science 2004;30:231e95. 2007;216:276e91.