Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 107 (2014) 398–414
www.elsevier.com/locate/solener

Potential improvements in the optical and thermal efficiencies

of parabolic trough concentrators
Men Wirz a, Jules Petit a, Andreas Haselbacher a, Aldo Steinfeld a,b,⇑
a
Department of Mechanical and Process Engineering, ETH Zurich, CH-8092 Zurich, Switzerland
b
Solar Technology Laboratory, Paul Scherrer Institute, CH-5232 Villigen, Switzerland

Received 7 October 2013; received in revised form 18 April 2014; accepted 4 May 2014
Available online 28 June 2014

Communicated by: Associate Editor Robert Pitz-Paal

Abstract

A detailed 3D heat transfer model is used to analyze the improvement potential of a typical parabolic trough concentrator (PTC)
system with step-wise idealizations of system components. Sigmoid functions are used to idealize and optimize the optical behavior
of the absorber tube selective coating. Reflectance, absorptance, and transmittance behavior is evaluated for the glass envelope, assessing
the potential performance of a system with an ideal selective glass. Optical properties of the primary concentrator mirror, such as reflec-
tivity as well as tracking and surface errors, are successively idealized to reveal the differences between ideal and real concentrators. In
addition, the effect of the supporting structure is analyzed by reducing the number of heat collection elements (HCE) per mirror module,
lowering the shaded area between HCEs, and removing the structure altogether. Applying these component idealizations individually
yielded increases in thermal efficiency between 4.3% and 7.3% compared to a selected benchmark PTC design, while a combination
of all idealizations resulted in an increase of 23%. The analysis of an alternative PTC design has shown that the idealization approach
is robust in that it predicts similar increases in thermal efficiency. In a second step, several secondary mirror designs are evaluated and
optimized, including a partially reflective glass surface, insulation in the vacuum annulus with reflective surfaces, as well as aplanatic
mirror and tailored secondary designs. The analysis of systems with secondary optics revealed potential increases in thermal efficiencies
between 0.8% and 1.6% compared to the benchmark design.
Ó 2014 Elsevier Ltd. All rights reserved.

Keywords: Parabolic trough; Efficiency; Optimization; Secondary concentrator

1. Introduction include the absorber tube, the protective glass envelope,

Parabolic trough concentrators (PTC) are currently the
most mature and cost-effective technology to generate elec-
tricity through concentrating solar power (CSP) (Price
et al., 2002). The main components of a PTC system
⇑ Corresponding author at: Department of Mechanical and Process
Engineering, ETH Zurich, CH-8092 Zurich, Switzerland. Tel.: +41 44 632
79 29.
E-mail address: aldo.steinfeld@ethz.ch (A. Steinfeld).
the reflective primary concentrator mirror, and the sup-
porting structure that connects and holds the receiver.
Determining the effect of the different components on the
system performance is required to assess the most fruitful
direction of development and optimization.
Previous studies focused on the effect of the receiver
geometry on the thermal efficiency of the PTC system by
varying the absorber-tube or glass-envelope diameters
(Kalogirou et al., 1994; Bakos et al., 2001; Forristall,
2003; Montes et al., 2010). Other studies analyze the effect

http://dx.doi.org/10.1016/j.solener.2014.05.002
0038-092X/Ó 2014 Elsevier Ltd. All rights reserved.
 M. Wirz et al. / Solar Energy 107 (2014) 398–414 399

Nomenclature

Abbreviations Q_ loss total heat loss of from the absorber tube (W)
CPC compound parabolic concentrator q_ 0loss total heat loss from the absorber tube per unit
CSP concentrating solar power length (W/m)
CSR circumsolar ratio s scaled distance between primary and secondary
DNI direct normal irradiance vertices (–)
FL focal line T temperature (K)
FV finite volume t thickness (m)
HCE heat collection element w aperture width (m)
HTF heat transfer fluid z vertical displacement from the FL (m)
IR infrared
MC Monte Carlo Greek symbols
PTC parabolic trough concentrator a absorptance (–)
RAI reflective annulus insulation b angular size of secondary objects (°)
RGS reflective glass surface e Emissivity (–)
gel efficiency of power block, generator, and pump
Symbols (%)
Aap total aperture area of the concentrator mirrors gth thermal efficiency of the PTC system (%)
(m2) h incidence angle at the concentrator aperture (°)
a1-4 coefficients (–) hs acceptance angle (°)
Ceff effective concentration at the absorber tube (–) k wavelength (lm)
Cs secondary concentration (–) kc cut-off wavelength (lm)
c coefficient (lm) q reflectance (–)
D diameter (m) s transmittance (–)
d distance (m) w rim angle (°)
e angular error (rad)
f aplanatic system focal length (m) Subscripts
IDNI direct normal irradiance (W/m2) a atmosphere
Ik solar spectral irradiance (W/m2lm) abs single absorber tube element
K scaled distance between secondary vertex and conc concentrator mirror
the FL (–) f heat transfer fluid
k thermal conductivity (W/mK) g glass envelope
l length (m) ins insulation material
NA scaled half-aperture of the primary mirror (–) sec secondary mirror
nHCE number of HCE per concentrator module (–) sh shaded end parts of the HCE
Ppump required power to pump the HTF (W) system solar field loop
Q_ abs solar power absorbed by the absorber tube (W) track tracking error
Q_ gain net heat gain of the HTF (W)

of various parameters of the primary concentrator mirror,
such as the rim angle, aperture width, mirror reflectivity,
slope error, tracking error, or mirror misalignment
(Bendt et al., 1979; O’Gallagher and Winston, 1988;
Kalogirou et al., 1994; Jenkins and Winston, 1996; Omer
and Infield, 2000; Forristall, 2003). To a lesser extent, anal-
yses have investigated the effect of operating conditions,
such as the direct normal irradiation (DNI), the solar inci-
dence angle, or the mass flow of the heat transfer fluid
(HTF) (Bakos et al., 2001; Forristall, 2003). In most stud-
ies, several parameters are analyzed individually, but rarely
are parameters varied simultaneously or are idealizations
implemented to quantify the overall improvement potential
of a certain component. To our knowledge, no studies have
addressed the effect of varying the optical properties of the
selective coating or the glass envelope or investigated the
impact of the supporting structure.
A different method to boost the overall performance of
CSP systems is to add a secondary mirror to increase the
concentration at the receiver. For PTC systems a number
of solutions have been proposed. Simple designs include
reflective surfaces on insulation material in the vacuum
annulus (Bakos et al., 2001) or shaped reflective surfaces
behind the absorber tube (McIntire, 1980; Winston, 1980;
Gee et al., 2002). Commonly implemented secondary
concentrators in point and line focus systems include the
400 M. Wirz et al. / Solar Energy 107 (2014) 398–414
compound parabolic concentrator (CPC) (Collares-
Pereira, 1979; Brunotte et al., 1996; Omer and Infield,
2000) and the hyperbolic concentrator, also known as the
“trumpet” (O’Gallagher and Winston, 1986; O’Gallagher
and Winston, 1988; Brunotte et al., 1996). CPCs and trum-
pets generally require lower primary mirror rim angles to
achieve the best optical performance. Alternative designs
have been proposed for larger rim angles, such as multiple
truncated CPCs (Collares-Pereira et al., 1991; Mills, 1995).
In addition, several aplanatic systems have been presented
(Gordon and Feuermann, 2005; Winston and Gordon,
2005; Gordon et al., 2008; Ostroumov et al., 2009;
Goldstein et al., 2011), for which the primary and second-
ary mirrors can be described by a parameterization
(Lynden-Bell, 2002). Besides analytical descriptions of the
geometries of the previously mentioned secondary designs,
tailoring methods using the edge-ray principle or “string”
method or the simultaneous multiple surface method are
used to construct secondary mirrors (Winston et al.,
2005; Chaves, 2008). A wide variety of such designs has
been proposed, such as a secondary shape that approxi-
mates a cone or V-shape concentrator (Friedman et al.,
1993; Gordon and Ries, 1993), designs that also allow a
sizeable gap between the secondary mirror and the absor-
ber (Benitez et al., 1997), a tailored secondary for conven-
tional PTC tubular receivers and a primary shape that is
approximately parabolic (Canavarro et al., 2013), and the
so-called “seagull”, named after the reflector’s resemblance
to a flying seabird (Ries and Spirkl, 1996; Spirkl et al.,
1997). This last design, which can be used with a conven-
tional PTC primary, consists of an analytically derived
involute and a tailored wing.
Several problems are commonly encountered when the
secondary mirror systems described above are applied to
PTCs. For example, when smaller rim angles of the
primary mirror are required, problems with tracking and
mechanical stability occur (Collares-Pereira et al., 1991;
Benitez et al., 1997; Omer and Infield, 2000). In addition,
some secondary-mirror designs are rather complex
(Collares-Pereira et al., 1991; Brunotte et al., 1996;
Chaves and Collares-Pereira, 2000), which may result in
high manufacturing and assembly costs. Several secondary
mirrors lead to increased heat losses by touching the hot
absorber tube, while others leave an intentional gap
between the mirror and the absorber and suffer from higher
optical losses (Collares-Pereira et al., 1991; Benitez et al.,
1997; Gordon and Feuermann, 2005; Canavarro et al.,
2013). From a modeling point of view, the proposed sec-
ondary designs are analyzed mostly in terms of their optical
performance alone. A combined optical and thermal anal-
ysis of a CSP system that incorporates secondary mirrors is
missing from the literature.
The objectives of this study are to assess the potential
improvements of the efficiency of PTC systems by compo-
nent idealizations and secondary optics, and to identify the
most promising avenues for further research and develop-
ment. Accordingly, the focus of this study is not on the
absolute performance of PTC systems, but on the differ-
ences in and the sensitivity of the optical and thermal
behavior relative to a generic PTC system due to changes
in its components. In contrast to previous investigations,
this study is based on a combined 3D optical and thermal
analysis. The combined analysis is necessary because the
optical performance of a PTC system has a strong impact
on its thermal performance. The approach adopted in this
study consists of two steps. In the first step, the optimiza-
tion potentials of major PTC system components, namely
the selective coating, the glass envelope, the primary con-
centrating mirror, and the supporting structure, are
assessed relative to a selected benchmark PTC system
design. The components are idealized step-by-step to eval-
uate the associated potential benefit. Starting from realistic
values for the components, the idealizations allow the sen-
sitivity and therefore the improvement potential of the
optical and thermal behavior to be quantified. In the sec-
ond step, selected secondary-mirror designs are subjected
to the same detailed 3D optical and thermal analysis. The
designs are selected to be compatible with conventional
PTC systems, which require a primary concentrator rim
angle of at least 80° and a tubular receiver, as well as over-
all simplicity to minimize manufacturing and assembly
costs. Among the selected designs are a simple reflective
glass surface, annulus insulation with reflective surfaces,
an aplanatic two-mirror configuration, and the tailored
“seagull” secondary. The secondary-mirror designs are
optimized for maximum thermal efficiency and compared
to the benchmark design. This detailed study enables an
evaluation of secondary mirrors from a thermal point of
view, extending previously published assessments on the
economic and optical value of these concepts.
2. Modeling
The optical and thermal performance of the PTC sys-
tems is evaluated using a detailed, steady-state, 3D heat
transfer model based on the Monte Carlo (MC) ray-tracing
and finite-volume (FV) methods. The following is only a
brief description of the heat transfer model. For more
detailed information, see (Wirz et al., 2012). The non-uni-
form distribution of incident solar radiation and the
exchange of thermally emitted radiation between the vari-
ous receiver surfaces and the surrounding atmosphere are
determined with an in-house MC ray tracing code
(Petrasch, 2010). The analysis is based on a geometric
model of a parabolic trough concentrator system. Struc-
tural elements such as support rods and shielding of recei-
ver joints are included in the geometric model to take into
account shading effects. The parabolic surface is modeled
as a specular reflector with reflectivity qconc and a Gaussian
standard deviation econc of the beam dispersion to repre-
sent angular reflection errors (Cooper and Steinfeld,
2011). Two circular cylinders with Fresnel reflection/refrac-
tion behavior are used to model the outer and inner surface
of the glass envelope, and an absorbing medium between
 M. Wirz et al. / Solar Energy 107 (2014) 398–414 401

the two cylinders models absorption in the glass. The with experimental data from both on-sun (Dudley et al.,
absorber tube is modeled as a cylinder with diffuse reflec- 1994) and off-sun tests (Burkholder and Kutscher, 2009).
tion and emission. For all surfaces, detailed spectral data The results showed that the model accurately predicts heat
for the optical properties may be used. The incident rays loss, thermal efficiency, and glass temperatures in PTC sys-
are directed at the parabolic concentrating mirrors at a tems (Wirz et al., 2012).
specified incidence angle using a CSR 5% sunshape angular Various performance parameters, such as the total heat
distribution (Buie et al., 2003) and a spectral distribution loss from the absorber tube and the optical and thermal
equal to the ASTM standard AM1.5 direct solar spectrum. efficiencies, are assessed for a typical high-temperature
To simulate tracking errors, the rays can be displaced PTC system located in Seville, Spain. Relevant system
around the concentrator’s axis by an angle etrack. parameters are summarized in Table 1. A single loop of
To account for the non-uniform radiation and tempera- the solar field consists of 48 concentrator modules in series
ture distribution around the circumference of the receiver that raise the temperature of the HTF from 290 °C at the
surfaces as well as the temperature changes along the recei- inlet of the loop to 550 °C at the outlet. The HTF is
ver’s axis, the heat transfer is evaluated at a number of sur- assumed to be a commercial molten salt (Coastal, 2004).
face segments. Therefore, the receiver surfaces are divided A current-generation receiver with an absorber tube outer
into 100 circumferential and 12 axial segments along the diameter of 0.07 m is selected (Archimede, 2013). The ther-
entire length lsystem of the solar field loop. On each axial mal conductivity of the absorber tube is given by kabs
segment, the HTF temperature is assumed to be constant. (T) = 0.0153T + 14.775 W/mK, while kg = 1.04 W/mK is
A 3D representation of the heat transfer in the PTC system used for the glass envelope (Forristall, 2003). Data of a
is generated by axial integration. The FV method is used to selective coating developed by ENEA is used to calculate
determine all modes of heat transfer between the HTF, the the spectral emissivity eabs of the absorber tube (Esposito
absorber tube and glass envelope surfaces, and the sur- et al., 2009). For the glass envelope, a two-band approach
rounding atmosphere. Specifically, convection from the with a cut-off wavelength of kc,g = 2.6 lm is used to model
absorber tube to the HTF and to the glass envelope across the optical properties. The glass transmittance sg is set to
the vacuum annulus, as well as convection from the glass to 0.965 for k 6 kc,g and 0.0 for k > kc,g, and its reflectance
the atmosphere is calculated using empirical correlations qg is set to 0.015 for k 6 kc,g and 0.11 for k > kc,g
(Wirz et al., 2012). In addition, conduction through the (Forristall, 2003; Burkholder and Kutscher, 2009). The
absorber and glass walls and through the supporting struc- optical parameters of the concentrator mirrors are set to
ture, as well as radiation from the glass to the atmosphere qconc = 0.92, econc = 4 mrad, and etrack = 1 mrad to pro-
is determined analytically (Wirz et al., 2012). All other duce realistic optical behavior. The selected parameters
radiation terms that are determined with the MC ray trac- yield results that agree well with measured incidence-
ing are introduced as heat sources to the FV algorithm. angle-dependent intercept factors (Riffelmann et al., 2006;
These radiation terms include the distribution of incident Schiricke et al., 2009) and optical efficiencies (Dudley
solar radiation absorbed by the absorber and in the glass et al., 1994; Maccari, 2006). The aperture width of the con-
envelope, and the thermally emitted radiation from the centrator is 6 m, which is a good approximation of cur-
absorber and the glass surfaces within the vacuum annulus rently used collectors (Fernández-Garcı́a et al., 2010).
that is either absorbed by these surfaces or passes through For all simulations presented in this study a solar incidence
the glass to the atmosphere. angle of h = 30.1° and a DNI of IDNI = 619 W/m2 is used.
The net heat gain of the HTF Q_ gain is equal to the con-
vective heat transfer from the absorber tube to the HTF.
The total heat loss of the solar field Q_ loss is determined by Table 1
the sum of the radiative heat transfer from the absorber Design parameters of a typical PTC system.
to the envelope and to the atmosphere, the convective heat Design parameter Symbol Value
transfer to the glass across the vacuum annulus, and the Direct normal irradiation IDNI 619 W/m2
conduction losses through the supporting structure (Wirz Solar incidence angle h 30.1°
et al., 2012). All results presented in this study have been Ambient temperature Ta 25 °C
HTF temperature Tf 290–550 °C
obtained with 2108 rays for the MC simulation of the inci-
Length of a solar field loop lsystem 584.64 m
dent solar radiation, while for the radiative exchange a Absorber tube; length labs 4.06 m
total of 2107 rays are used for each axial integration step. Absorber tube; outer diameter Dabs 0.07 m
This results in an RMS temperature difference of 0.05 K or Absorber tube; wall thickness tabs 0.003 m
less between iterations at convergence for all surface seg- Glass envelope; outer diameter Dg 0.125 m
Glass envelope; wall thickness tg 0.003 m
ments and percentage efficiencies that are converged to
Concentrator; length lconc 12 m
one decimal figure. Literature data is used for the material Concentrator; aperture width wconc 6m
properties of the absorber and the glass (Forristall, 2003) Concentrator; rim angle wconc 80°
and air (Lienhard and Lienhard, 2011). This heat transfer Concentrator; reflectivity qconc 0.92
model was validated in a previous publication (Wirz Concentrator; angular reflection error econc 0.004 rad
Concentrator; tracking error etrack 0.001 rad
et al., 2012), in which simulation results were compared
402 M. Wirz et al. / Solar Energy 107 (2014) 398–414
These operating conditions, determined through weighted
yearly average values, represent a design point that pro-
duces thermal efficiencies that are lower than peak values
and are more similar to the expected yearly average
performance (Wirz et al., 2014). The optical efficiency of
the PTC system is defined as the solar power absorbed
by the absorber tube divided by the DNI over the total
aperture area of the concentrator mirrors,
Q_ abs
gopt ¼
I DNI  Aap
The secondary mirror surfaces described here are
modeled as perfect specular reflectors with a reflectivity
of qsec = 0.95. Emission of radiation and convection from
the secondary surface, as well as thermal conduction
through the mirrors are neglected apart from the insulation
material discussed in Section 3.3.2. The thermal efficiency
of the PTC system is defined as the net useable thermal
power (i.e., the net heat gain of the HTF minus the power
required to pump the HTF) divided by the DNI over the
total aperture area,
Q_ gain  P pump =gel Q_ abs  Q_ loss  P pump =gel
gth ¼ ¼ ð2Þ
I DNI  Aap I DNI  Aap
where Ppump is the required pumping power for fully-
developed flow (Incropera et al., 2006) (determined using
a friction-factor correlation (Serghides, 1984)), and
gel = 32.7% is the product of typical efficiencies of the
power block, the electrical generator, and the HTF pump
(Patnode, 2006; Manzolini et al., 2011). The required
pumping power is included in the calculation of the ther-
mal efficiency to take into account larger pressure losses
caused by smaller absorber tube diameters. These pressure
losses become significant at absorber diameters below
about 0.04 m. This ensures that the outcome of the optimi-
zation process is not biased towards smaller absorber
diameters.
3. Results
3.1. Benchmark design
The assessment of potential improvements starts with
the typical PTC system described in Section 2 and Table 1.
At this stage, the optimization does not yet consider any
changes in material properties or the addition of secondary
optics. Instead, the optimization considers only the absor-
ber tube diameter Dabs and the concentrator rim angle
wconc. The concentrator aperture width is not changed to
keep the theoretical solar input at the system aperture
constant. Fig. 1 shows a fitted contour plot of the thermal
efficiency as a function of the two parameters. The contour
plot is obtained from a fourth-order polynomial least-
squares fit to results from 54 simulations with the detailed
3D heat transfer model. (The RMS difference between the
fit and the results from the heat transfer model is around
ð1Þ
Fig. 1. Contour plot of thermal efficiency as a function of absorber tube
diameter and concentrator rim angle fitted to a total of 54 simulation
points (scatter dots). The optimum configuration with Dabs = 0.058 m and
wconc = 112° is indicated by the circle.
0.02 percentage points.) In the figure, the global maximum
of gth = 60.0% is indicated by the circle. Values of the
thermal efficiency, together with the optical efficiency and
the heat loss from the absorber tube per unit length of
the receiver, are listed in Table 2.
The relative contribution of the factors that affect the
thermal efficiency are plotted in Fig. 2 as a function of Dabs
for wconc = 110°. For commonly used absorber tube
diameters (e.g., Dabs = 0.07 m), the pumping losses are
negligible. For smaller diameters, however, the pumping
losses increase significantly, reaching values of 0.3% of
the solar input IDNIAAp at Dabs = 0.04 m. With larger
absorber diameters the thermal losses increase mostly due
to the larger surface area of the absorber tube, accounting
for 4.2–8.4% of the solar input. The optical losses increase
with smaller absorber sizes and represent 32.7–37.5% of the
solar input. Similar trends and relative contributions are
observed for these parameters at concentrator rim angles
other than wconc = 110°. The thermal efficiency is observed
to exhibit a local maximum because of the opposite trends
of the optical and heat losses.
From Fig. 1, it can be seen that the maximum thermal
efficiency is achieved at rim angles around 110°, depending
on the chosen absorber diameter. For larger absorber sizes,
the optimum rim angle increases, reaching a global maxi-
mum at wconc = 112° and Dabs = 0.058 m. As the absorber
diameter is not selected based on a specific sun angle, the
mean distance of the receiver from the primary mirror
has a stronger effect on the final concentration. In turn, this
has a significant effect on the optical efficiency of the recei-
ver. Therefore, the optimum for the rim angle moves from
90°, where the highest theoretical concentration at 100%
interception occurs, towards 120°, where the smallest mean
 M. Wirz et al. / Solar Energy 107 (2014) 398–414 403

Table 2
Results of a typical PTC system, the optimum configuration, and the selected benchmark design.
Configuration Dabs (m) wconc (°) gopt (%) q_ 0loss (W/m) gth (%)
Typical PTC system 0.070 80 75.1 268 58.6
Optimum design 0.058 112 66.0 219 60.0
Benchmark design 0.060 80 65.0 229 58.8

following, the optimization potential of the benchmark

design is assessed by idealizing selected components. In
Section 3.2, the absorber tube diameter is kept constant
at Dabs = 0.06 m, whereas in Section 3.3, the absorber size
is varied as part of the optimization analysis.

3.2. Optimization potential of idealized PTC components

3.2.1. Absorber tube selective coating

The absorber tube selective coating ensures high solar
absorption and low re-radiation losses. To assess the opti-
mization potential of the selective coating, the spectral
behavior can be idealized. One way of modeling the
spectral properties of selective coatings, for example the
spectral reflectivity, is by using logistic functions such as
sigmoid curves,
a1
Fig. 2. Main factors impacting thermal efficiency, including optical qabs ðkÞ ¼ þ a2 ð3Þ
efficiency, heat losses, and the required pumping power, plotted as a 1 þ expða3 =k þ a4 Þ
function of Dabs for wconc = 110°. Heat losses and pumping are shown as a
fraction of the solar input identical to how the optical and thermal To study the effect of the spectral behavior of selective
efficiencies are defined. coatings, a sigmoid function is fitted to the spectral reflec-
tance data of the ENEA coating used in this study
(Esposito et al., 2009) by determining appropriate coeffi-
distance between receiver and primary concentrator is cients: a1 = 0.992, a2 = 0.0105, a3 = 9.302 lm, and
attained. a4 = 4.559. Fig. 3 shows the spectral reflectance curve
However, rim angles above 90° have several mechanical of the ENEA coating (solid line) and the fitted sigmoid
and economic disadvantages. A mirror with a rim angle of curve (dashed line). Simulations with the benchmark
110° has a glass surface that is over 15% larger compared to
a typical concentrator mirror with a rim angle of 80°,
increasing material cost and requiring a heavier supporting
structure. The larger mirror surface also causes greater
wind loads and the associated deformation of the mirror
surface may negate the improved optical performance. In
the following analyses, a rim angle of 80° is used as baseline
since it combines good optical performance and a compact
mirror-receiver configuration. Since it is also a configura-
tion widely used today, a receiver optimized for this rim
angle can be combined with a commercially available
mirror. The optimum absorber diameter for this baseline
rim angle of 80° is around 0.06 m. This absorber size
produces near-optimum efficiencies for virtually all investi-
gated rim angles, as the optimum range is very broad (see
Fig. 1). Therefore, these geometric characteristics define
what is called the benchmark design in this article. The
benchmark design is identical to the typical PTC system
described in Section 2 and in Table 1, with the exception
of Dabs = 0.06 m. The thermal efficiency of the benchmark
design is gth = 58.8%. All subsequent analyses in this Fig. 3. Spectral reflectance for the ENEA coating, the fitted sigmoid
study will be compared to this benchmark design. In the curve, and two generic sigmoid curves.
404 M. Wirz et al. / Solar Energy 107 (2014) 398–414

Table 3
Simulation results for different spectral data of the selective coating.
Configuration gopt (%) q_ 0loss (W/m) gth (%) Dgth (%)
Benchmark design (ENEA coating) 65.0 229 58.8 –
Sigmoid fit to ENEA coating 65.0 228 58.8 0.0
Sigmoid fit; no solar reflection 65.7 230 59.4 0.6
Sigmoid fit; no IR emission 65.0 195 59.6 0.8
Sigmoid fit; no solar refl./IR emiss. 65.6 197 60.2 1.4
Sigmoid A; c = 10 lm, kc,abs = 1.7 lm 63.8 99 61.1 2.3
Sigmoid B; c = 1000 lm, kc,abs = 2.3 lm 68.1 82 65.8 7.0

design, using the fitted sigmoid curve to calculate the spec-

tral emissivity of the absorber tube, yielded results very
close to those obtained using the actual spectral data, as
shown in Table 3. The sigmoid curve can now be used to
easily implement idealizations. Perfect solar absorption is
attained by setting a1 = 1.003 and a2 = 0, yielding qabs = 0
for the wavelength limit below the inflection point, here
referred to as the cut-off. Setting a1 = 1.006 and
a2 = 0.0105 produces a sigmoid curve with ideal reflectance
at higher wavelengths, resulting in qabs = 1 for wavelengths
above the cut-off. Finally, using a1 = 1.017 and a2 = 0 com-
bines both effects, yielding a sigmoid curve with qabs = 0
below and qabs = 1 above the cut-off. The improvements
in heat loss and thermal efficiency for each of these ideali-
zation steps are listed in Table 3. Starting from a realistic
spectral behavior, such as the sigmoid fit of the ENEA
coating, assuming perfect absorption characteristics in the
solar spectrum allows for improvements in thermal effi-
ciency of 0.6% due to improved optical efficiency. An Fig. 4. Heat loss as a function of the cut-off wavelength for different
increase of around 0.8%, achieved through lower heat sigmoid curve coefficients.
losses, is possible with a coating that has zero thermal
emission at higher wavelengths. Consequently, a coating
displaying both ideal behaviors allows for an increase in
the thermal efficiency of 1.4%.
In addition to idealizing the levels of reflectance at both
ends of the wavelength spectrum, the actual shape of the
spectral reflectance curve can also be optimized. For this
purpose, a simpler sigmoid function is used, assuming ideal
behavior in the solar part of the spectrum (qabs = 0) and
the infrared (IR) part of the spectrum (qabs = 1),
1
qabs ðkÞ ¼ ð4Þ
1 þ expðc=k  c=kc;abs Þ
where c is a coefficient determining the shape of the curve
and kc,abs is the cut-off wavelength. The higher c, the more
the reflectance curve approaches a step function. Fig. 3
shows the spectral reflectance of two examples of such sim-
plified sigmoid curves. The curve with coefficients
c = 10 lm, kc,abs = 1.7 lm (“Sigmoid A”) is a curve similar
to the sigmoid fit of the ENEA coating, while the example Fig. 5. Optical efficiency as a function of the cut-off wavelength for
with c = 1000 lm, kc,abs = 2.3 lm (“Sigmoid B”) approxi- different sigmoid curve coefficients, together with the ASTM AM1.5D
mates a step function with a cut-off at kc,abs. Results of sim- solar spectrum (right axis).
ulations performed with various simplified sigmoid curves
are presented in the following figures. Selected results of loop is shown as a function of the cut-off wavelength kc,abs
configurations are summarized in the last two rows of for several values of the coefficient c. In subsequent plots, a
Table 3. In Fig. 4, heat loss per unit length of the solar field spline is fit through the data points. As the cut-off
 M. Wirz et al. / Solar Energy 107 (2014) 398–414
wavelength increases, the heat loss increases because the
total emissivity becomes larger. When c is increased, the
sigmoid curve approaches a step function, resulting in
lower total emissivities for a specific cut-off wavelength.
The cut-off wavelength also has a significant effect on the
optical efficiency, as can be seen from Fig. 5. The optical
efficiency increases for larger kc,abs because of greater total
solar-weighted absorptance. With increased c, however, a
break in the continuous rise of the optical efficiency is
observed between cut-off wavelengths of about 1.8–
2.0 lm. This break stems from the drop in solar irradiance
Ik for this range of wavelengths, as can be seen from the
ASTM AM1.5D solar spectrum that is also shown in
Fig. 5. Between wavelengths of around 2.0 and 2.5 lm,
the optical efficiency increases again. Above 2.5 lm, the
solar intensity becomes extremely small, and the optical
efficiency reaches a maximum that stays virtually constant
for all higher kc,abs.
Because of the shape of the solar spectrum, the optical
efficiency is relatively insensitive to a cut-off wavelength
in the range of 1.8–1.9 lm, provided that the reflectance
curve approaches a step function. However, the thermal
efficiency is affected since the heat losses increase steadily
with increasing cut-off wavelength, see Fig. 4. This may
be seen from Fig. 6, where the thermal efficiency is plotted
as a function of the cut-off wavelength. For higher c, the
efficiency drops in the spectral band between 1.8 and
2.0 lm due to the reduced solar input. Because of the
increased heat losses in this part of the spectrum, the
maximum thermal efficiency is reached at cut-off wave-
lengths that are below those determined for the best optical
efficiency. To illustrate this point, the optimum cut-off for
different sigmoid coefficients c is plotted in Fig. 7. For
c = 1000 lm, the optimum cut-off is around kc,abs = 2.3
lm, at which a thermal efficiency of gth = 65.8% is reached
Fig. 6. Thermal efficiency as a function of the cut-off wavelength for
different sigmoid curve coefficients.
405
Fig. 7. Optimum cut-off wavelength as a function of the sigmoid
coefficient c.
(the highest of any of the configurations evaluated here).
For smaller c, the optimum kc,abs decreases significantly.
The optimum cut-off for a more realistic approximation
of the reflectance, such as a sigmoid function with
c = 10 lm, is located at a wavelength of around 1.7 lm.
As a result, the optimum cut-off wavelength of a selective
coating varies significantly, depending on the shape of the
spectral reflectance curve. In conclusion, an ideal selective
coating, displaying a spectral step function behavior with
a cut-off wavelength of kc,abs = 2.3 lm and reflectance
values of qabs = 0 for k < kc,abs and qabs = 1 for k > kc,abs,
has an improvement potential of around 7% compared to
the benchmark design that uses an existing selective
coating.
3.2.2. Glass envelope
The optimization potential of the glass envelope is deter-
mined by solely changing the values below and above the
cut-off wavelength kc,g = 2.6 lm. Three idealizations are
assumed that are analyzed both individually and in combi-
nation. Ideal behavior in the solar spectrum is achieved by
setting qg = 0 and/or ag = 0 for k 6 kc,g. This measure pri-
marily increases the optical efficiency. In the IR spectrum,
i.e., above the cut-off, ideal behavior is achieved by assum-
ing zero emission from and zero absorption by the glass,
reducing the radiative heat losses from the absorber tube.
Table 4 shows the results of the corresponding simulations.
It can be seen that compared to the benchmark design, the
largest improvements for a single measure is achieved by
setting the IR absorption to zero, significantly reducing
the heat losses from the absorber tube. Therefore, an ideal
selectively reflective coating in the IR spectrum could
improve the overall efficiency by around 4.2%. Comparable
improvements in the solar spectrum can only be achieved
by combining the effects of a perfect anti-reflective coating
406 M. Wirz et al. / Solar Energy 107 (2014) 398–414

Table 4
Simulation results for various configurations based on idealizing the optical properties of the glass envelope. The notation “x/y” indicates that the optical
property is equal to x below the cut-off and y above it.
Configuration qg (%) sg (%) ag (%) gopt (%) q_ 0loss W/m gth (%) Dgth (%)
Benchmark design 1.5/11 96.5/0 2/89 65.0 229 58.8 -
No solar reflection 0/11 98/0 2/89 66.5 229 60.2 1.4
No solar absorption 1.5/11 98.5/0 0/89 66.5 229 60.2 1.4
No IR absorption 1.5/100 96.5/0 2/0 65.0 74 63.0 4.2
No solar refl./abs. 0/11 100/0 0/89 68.1 230 61.8 3.0
No solar refl./IR abs. 0/100 98/0 2/0 66.5 75 64.4 5.6
No solar/IR abs. 1.5/100 98.5/0 0/0 66.5 74 64.5 5.7
No solar/IR refl./abs. 0/100 100/0 0/0 68.2 75 66.1 7.3

and perfect transmittance behavior, for example by using a

very thin envelope. By removing both reflection and
absorption losses from the glass in the solar spectrum, an
increase in optical efficiency, and consequently also in the
thermal efficiency, of 3% with respect to the benchmark
design can be achieved. A glass envelope with completely
ideal optical properties and a cut-off wavelength of
2.6 lm has a total improvement potential of around 7.3%
compared to the benchmark design.

3.2.3. Primary concentrator mirror

The optimization potential of the primary concentrator
is determined by idealizing the optical properties of the
mirror. The parameters include the reflectivity, the angular
reflection error, and the tracking error. The results of the
corresponding simulations are summarized in Table 5.
Increasing the reflectivity from the benchmark value of
qconc = 0.92 to 1.0 assumes a mirror with an ideal reflection
material without any absorption losses. This measure line- Fig. 8. Thermal efficiency for qconc = 0.92 as a function of the angular
arly increases the solar input to the system, resulting in a reflection error for two different tracking errors.
linear increase in gth. From Table 5 it can be seen that with
this increase in reflectivity, a gain in thermal efficiency of
around 5.5% is achieved, which is roughly equal to the achieved. When the reflection error approaches zero, the
increase in reflected solar power times the optical efficiency. optical efficiency reaches its maximum value of
The effects of the angular reflection error econc, simulat- gopt = 65.2%, being unaffected by the tracking error, as
ing dispersion caused by a non-ideal mirror surface, are virtually all the radiation is intercepted by the absorber
visualized in Fig. 8 for the cases of etrack = 0 and 1 mrad (see Table 5). The tracking error has an increasingly
and qconc = 0.92. For the case of etrack = 1 mrad, as the dis- significant effect only for larger econc, causing a reduction
persion increases, the optical efficiency decreases, leading of 0.1% in thermal efficiency for the benchmark design with
to a considerable drop in thermal efficiency for econc > 2 econc = 4 mrad and qconc = 0.92. While the mirror reflectiv-
mrad. Assuming a perfect mirror surface without any ity has by far the largest individual impact on the thermal
surface error, an improvement in thermal efficiency of efficiency, combining all three idealizations leads to an
around 1.2% compared to the benchmark design can be improvement of around 6.8%.

Table 5
Simulation results for different configurations for various idealizations of the primary concentrator parameters.
Configuration etrack (mrad) econc (mrad) qconc (–) gopt (%) q_ 0loss (W/m) gth (%) Dgth (%)
Benchmark design 1.0 4.0 0.92 65.0 229 58.8 –
No tracking error 0.0 4.0 0.92 65.2 229 58.9 0.1
No surface error 1.0 0.0 0.92 66.2 229 60.0 1.2
No mirror absorption 1.0 4.0 1.00 70.6 228 64.3 5.5
No track./surf. 0.0 0.0 0.92 66.2 229 60.0 1.2
No track./abs. 0.0 4.0 1.00 70.8 228 64.5 5.7
No surf./abs. 1.0 0.0 1.00 71.9 228 65.6 6.8
No track./surf./abs. 0.0 0.0 1.00 71.9 228 65.6 6.8
 M. Wirz et al. / Solar Energy 107 (2014) 398–414
Fig. 9. Schematic of the supporting structure.
3.2.4. Supporting structure
Another source of losses is the shading of the absorber
tube by the structural components of the system, such as
the support arms or the end bellows of the receivers.
Fig. 9 shows a side view of the supporting structure as it
is used in the MC ray-tracing simulations and the heat-
transfer analysis, as well as the relevant geometrical param-
eters. Three measures are analyzed to determine the impact
of reducing shading losses. In a first step, the number of
heat collection units (HCE) per concentrator module
(nHCE) is reduced. In the benchmark design, this value is
nHCE = 3, with labs = 4.06 m. In the optimization, nHCE,
and with it the number of support arms, is reduced to
two and one, respectively. The length of the concentrator
module (lconc = 12 m) and the distance between two
modules (dconc = 0.18 m) are kept constant, resulting in
an increase of labs to 6.09 m and 12.18 m, respectively.
The length of the shaded end parts of the HCE (dsh) is
coupled to dconc for practical reasons to fulfill dconc = dsh
for all configurations. As a second measure, dsh and dconc
are reduced from 0.18 m in the benchmark design to zero.
The configurations with dconc = dsh = 0 simulate an ideal
HCE without any bellows or connection parts at the end,
which shade the absorber tube and decrease the useful
length of the tube. With this idealization, the receiver
becomes one single absorber tube and glass envelope
running along the entire length of the solar field, with
one single concentrator mirror of the same length. Finally,
simulations are performed with and without support arms,
to determine the associated shading losses.
In Table 6, the results of selected simulations are shown
that implemented these idealizations. The largest effect in
terms of improved thermal efficiency is achieved when
reducing the number of HCE per concentrator module,
which requires longer receivers and fewer support arms.
Reducing the number of HCE per module to two and
one leads to increases in the thermal efficiency of 1.4%
and 2.8%, respectively. The increases in thermal efficiency
are achieved both through higher optical efficiency but also
through lower heat losses. Assuming an ideal receiver
whose entire surface is irradiated and that requires no
end parts to connect the single HCE (dsh = 0), an addi-
tional improvement of 0.8% is possible with respect to
the case of using 1 HCE per module. The higher efficiency
407
is achieved over a shorter loop length, as the distance
between the concentrator modules is also reduced to
dconc = 0. A completely idealized structure system, where
support arms are not required, raises the thermal efficiency
by another 0.6%, resulting in a total improvement of 4.3%
compared to the benchmark design.
3.3. Secondary optics
Most improvements discussed in the previous sections
depend on material properties that may be difficult to
attain in practice. Nevertheless, the results are useful in
guiding and prioritizing research and development efforts.
The use of secondary optics at the receiver is another
option to increase the optical efficiency and reduce heat
losses by adapting the geometry of the absorber tube. In
the following, four types of secondary-mirror designs are
analyzed both optically and thermally and optimized for
maximum thermal efficiency. The designs include a reflec-
tive glass surface, annulus insulation material with reflec-
tive surfaces, an aplanatic two-mirror configuration, and
a tailored “seagull” secondary concentrator. The bench-
mark design is again used as a starting point for the imple-
mentation of secondary optics, but in contrast to the
previous section, the absorber tube diameter is now consid-
ered to be variable in the optimization process.
3.3.1. Reflective glass surface
The first type of secondary optics to be investigated is a
simple reflective glass surface (RGS) (Benitez et al., 1997;
Gee et al., 2002). This entails coating a portion of the inner
surface of the glass envelope with a highly reflective mate-
rial (qsec = 0.95). In addition, both the absorber tube and
the glass envelope are vertically displaced from the focal
line (FL) of the primary mirror. The absorber is moved
downwards to intercept radiation that would pass beneath
the absorber. The envelope is moved upwards so that the
radiation missing the top of the absorber is reflected back
towards it by the RGS. The reflective surface also reduces
heat losses by reflecting thermally emitted radiation from
the hot absorber tube. Fig. 10 shows a schematic of this
RGS system, together with selected rays from the MC sim-
ulation of the incident concentrated solar radiation. The
main parameters affecting the performance of the RGS
408 M. Wirz et al. / Solar Energy 107 (2014) 398–414

Table 6
Simulation results of configurations for various idealizations of the supporting structure.
Configuration nHCE (–) dsh (m) gopt (%) q_ 0loss (W/m) gth (%) Dgth (%)
Benchmark design 3 0.18 65.0 229 58.8 –
2 HCE per module 2 0.18 66.4 225 60.2 1.4
1 HCE per module 1 0.18 67.7 221 61.6 2.8
1 HCE, no spacing 1 0 68.4 221 62.4 3.6
1 HCE, no spacing/support 1 0 69.0 218 63.1 4.3

to the absorber tube. In addition, it should be noted that

the optical efficiency of a conventional system with
Dabs = 0.050 m and without an RGS would be around
63.1%. Therefore, although the secondary mirror shades
most of the receiver, the RGS reflects enough incident solar
radiation to achieve an increase in gopt. As a consequence,
the effective concentration at the absorber tube, defined by
Ceff = goptwconc/pDabs, is 24.7 for the optimum RGS system
as compared to Ceff = 20.7 for the benchmark design.

3.3.2. Reflective annulus insulation

Fig. 10. Visualization of the optimum configuration using a reflective
glass surface (RGS) on the inside of the envelope, together with selected
rays and design parameters. FL denotes the focal line.
configuration are the absorber diameter Dabs, the displace-
ment of the absorber zabs, the displacement of the envelope
zg, and the length of the reflective surface specified by the
angle bg. An initial analysis showed that the effect of the
envelope diameter is not that significant (with or without
RGS), which is consistent with previous observations
(Forristall, 2003). Therefore, the benchmark value
Dg = 0.125 m was used for all the simulations discussed
below.
Parameter studies yielded maximum thermal efficiency
with Dabs = 0.050 m, zabs = 10 mm, zg = 14 mm, and
bg = 83°. The optimum configuration using an RGS with
the above parameters is shown in Fig. 10. Detailed results
of the optimization of all secondary optics concepts dis-
cussed in this chapter are summarized in Table 7. It can
be seen that with an RGS a small decrease in optical effi-
ciency occurs due to the smaller absorber. Also, the heat
losses are reduced by about 24%, leading to an improve-
ment in the thermal efficiency of 1% compared to the
benchmark design. The reduction in heat loss is larger than
that caused by the reduction in the absorber surface, so a
portion of the thermally emitted radiation is reflected back
The next configuration studied is the reflective annulus
insulation (RAI), in which the upper half of the vacuum
annulus is be filled with an insulation material (kins = 0.04 -
W/mK). This eliminates radiation losses from the absorber
tube in that area, but also increases the conductive heat
transfer from the absorber to the glass envelope. In addi-
tion, the exposed surfaces of the insulation material in
the vacuum annulus are assumed to be coated with a highly
reflective material (qsec = 0.95) to allow a portion of the
solar radiation that hits the insulation to be reflected back
to the absorber tube. The configuration resembles previ-
ously presented designs (Bakos et al., 2001; Al-Ansary
and Zeitoun, 2013), although the configuration is modified
in this study to allow for more parameters to be optimized.
Similar to the RGS, the parameters varied in the optimiza-
tion study of the RAI are Dabs, zabs and zg, as well as the
dimensions of the insulation, determined by the angles babs
and bg, see Fig. 11.
The parameter studies revealed that the best perfor-
mance is achieved when the RAI covers less than the entire
upper half of the absorber tube, as is the case in previously
proposed designs (Bakos et al., 2001; Al-Ansary and
Zeitoun, 2013). The lower optical efficiency due to the
blocked incident solar radiation of such a system could
not be compensated by the lower heat losses that would
occur as more of the hot absorber tube surface is covered
by the insulation material. Instead, the insulation is best

Table 7
Simulation results of different optimum configurations, using secondary optics.
Configuration Dabs (m) gopt (%) Ceff (–) q_ 0loss (W/m) gth (%) Dgth (%)
Benchmark design 0.060 65.0 20.7 229 58.8 –
RGS 0.050 64.7 24.7 175 59.8 1.0
RAI 0.052 64.6 24.0 178 59.7 0.9
Aplanat 0.048 65.1 25.9 166 60.4 1.6
Seagull 0.046 64.4 26.6 173 59.6 0.8
 M. Wirz et al. / Solar Energy 107 (2014) 398–414
Fig. 11. Visualization of the optimum configuration using reflective
annulus insulation (RAI, indicated by the gray shading) between the
absorber and the envelope, together with selected rays and design
parameters.
limited to the uppermost part of the absorber tube where
only little concentrated radiation is incident. The optimum
configuration is displayed in Fig. 11 and is achieved with
Dabs = 0.052 m, zabs = 9 mm, zg = 11 mm, babs = 43°,
and bg = 83°. With this configuration, the heat losses are
reduced by 51 W/m compared to the benchmark design,
see Table 7. Despite the lower optical efficiency, an increase
in the thermal efficiency of around 0.9% is achieved. The
improvements obtained with the RAI are hence slightly
lower than with the RGS.
3.3.3. Aplanatic mirrors
In addition to the two simple designs assessed in the pre-
vious sections, more sophisticated secondary designs can be
considered. Aplanatic mirrors are a promising option
because they allow designs close to the thermodynamic
limit (Gordon and Feuermann, 2005). However, most of
the proposed designs are better suited for flat absorbers,
and only a few configurations come into consideration
for the use of tubular absorbers in PTC systems. The geom-
etries of both the primary and secondary mirrors are con-
structed using an analytic solution of the mirror contours
with parameters s, K, and NA (Ostroumov et al., 2009).
The parameter s is the distance between the vertices of
the primary and the secondary mirror, divided by the sys-
tem’s focal length f. The distance between the vertex of
the secondary and the FL, divided by f, is defined as K.
The numerical aperture NA is a measure of the maximum
attainable concentration and the maximum angle at which
a ray reaches the focus. The value of NA is also equal to
half of the primary’s aperture width, divided by f. In this
study, the aperture width wconc = 6 m is set as a constraint
for any configuration, resulting in a focal length of
f = wconc/2NA (Ostroumov et al., 2009). For the primary
mirror, the same optical properties (emissivity, angular
errors) apply as for the parabolic primary of the bench-
mark design, while for the secondary mirror qsec = 0.95 is
used. Of all the analyzed configurations, only the category
of dual-mirror aplanats with s > 0 and K > 0 yielded results
409
that are comparable to the performance of the benchmark
design. Some configurations of other categories (e.g., s < 0,
K < 0) could not match the benchmark design as the larger
secondary shapes resulted in low optical efficiencies. Other
designs were dismissed because they either lack acceptable
compactness or are unsuitable for circular receivers. In the
optimization study, Dabs, s, K, and NA are simultaneously
varied to find the configuration with the maximum thermal
efficiency. For any given configuration, the vertical position
of the absorber tube is such that the top of the tube touches
the vertex of the secondary mirror, see Fig. 12.
The optimization study indicates that the maximum
thermal efficiency of 60.4% is achieved with Dabs = 0.048
m, s = 0.59, K = 0.0015, and NA = 0.998. The correspond-
ing receiver and secondary mirror are shown in Fig. 12.
Results for the optimum configuration, such as the optical
and thermal efficiency, the effective concentration, and the
total heat loss, are listed in Table 7. It can be seen that the
thermal efficiency of the optimized aplanatic configuration
is 1.6% higher than that of the benchmark design. This is
partly because of a slightly higher optical efficiency despite
the rather large shaded area above and beside the absorber
tube. It is also due to lower heat losses, due to the smaller
absorber size and the thermally emitted radiation that is
reflected back to the absorber. The optimum configuration
also exhibits a similar compactness than the benchmark
design, corresponding to a parabolic trough with a rim
angle of approximately 81°. A drawback of this design is
that it would be difficult to implement in practice because
the secondary mirror intersects the glass envelope. One
solution would be to increase the diameter of the glass
envelope to allow the entire secondary mirror to fit inside.
That way, the secondary would lie completely within the
protective vacuum atmosphere, allowing for the use of bet-
ter suited materials (Collares-Pereira et al., 1991; Benitez
et al., 1997). Furthermore, higher heat losses are incurred
since the secondary mirror touches the absorber tube.
(These losses were not considered in this study). A small
gap between the secondary and the absorber may reduce
heat loss with a slight reduction in the optical efficiency.
Fig. 12. Visualization of the optimum configuration using aplanatic
mirrors, together with selected rays.
410 M. Wirz et al. / Solar Energy 107 (2014) 398–414
3.3.4. Tailored seagull
The final secondary design is the tailored seagull concen-
trator. The mirror shape is constructed according to (Ries
and Spirkl, 1996) using two main parameters, hs and Cs.
Normally, hs is the solar half-angle according to which
the concentrator is tailored. In this case, however, the
spread of the incident solar radiation follows a Gaussian
distribution due to the angular reflection error of the con-
centrator mirror (Cooper and Steinfeld, 2011). As a result,
hs can be interpreted as the variable acceptance angle, since
a secondary that is tailored using hs accepts only the rays
that are within this angular spread. The final size of the
absorber tube is then determined using the desired second-
ary concentration Cs, according to (Ries and Spirkl, 1996),
wconc sin hs
Dabs ¼
sin wconc C s
In this study, the wing of the secondary reflector is
designed as a straight line, which is a good approximation
of the tailored shape (Spirkl et al., 1997). Fig. 13 shows a
visualization of such a tailored seagull design. In the opti-
mization study, hs and Cs are simultaneously varied to find
the configuration with the maximum thermal efficiency.
The results of the optimization study are presented as a
contour plot in Fig. 14, where the thermal efficiency is
shown as a function of hs and Cs. The contour plot is
obtained from a fourth-order polynomial least-squares fit
to results from 90 simulations with the detailed 3D heat
transfer model. (The RMS difference between the fit and
the results from the heat transfer model is around 0.07 per-
centage points.) Lowering hs decreases the absorber size,
leading to worse interception and reducing the optical effi-
ciency. Increasing Cs further decreases the diameter of the
absorber tube and also increases the width of the tailored
wing of the secondary, causing additional shading. Even
with reduced heat losses from the smaller absorber, the
lower optical performance leads to lower thermal effi-
ciency. On the other hand, an increase in the acceptance
angle causes better optical performance due to the
increased absorber size, especially for small Cs, but due
to the higher heat losses, these configurations again
Fig. 13. Visualization of the optimum configuration using the tailored
seagull secondary design, together with selected rays.
ð5Þ
Fig. 14. Contour plot of thermal efficiency as a function of the acceptance
angle hs and secondary concentration Cs fitted to a total of 90 simulation
points (scatter dots).
decrease the thermal efficiency. The maximum performance
with gth = 59.6% is achieved with the configuration dis-
played in Fig. 13, using hs = 13.1 mrad and Cs = 1.72,
resulting in an absorber diameter of Dabs = 0.046 m. Val-
ues for the optical and thermal efficiencies, effective concen-
tration, and total thermal losses are listed in Table 7. The
optical efficiency of this optimum configuration is lower
than that of the benchmark design due to the small absor-
ber size. However, this is outweighed by the lower heat
losses, leading to an overall increase in thermal efficiency
of around 0.8%. Similar to the optimized aplanatic config-
uration, this secondary mirror design is difficult to imple-
ment in practice because the secondary touches the
absorber. Again, a small gap between the absorber tube
and the secondary mirror might improve the practicality
and the overall performance of this design at the expense
of reduced optical efficiency.
3.3.5. Temperature distribution
Besides raising the thermal efficiency, some of the sec-
ondary designs presented here have the additional benefit
of a more uniform distribution of the incident concentrated
solar radiation. This leads to smaller temperature gradients
along the circumference of the absorber tube, thereby
reducing heat losses and thermal stresses. The non-uniform
distribution of both the incident solar radiation and surface
temperatures can be accurately determined with the 3D
heat transfer model. In Fig. 15, the temperature distribu-
tion on the absorber tube is plotted as a function of the
circumferential angle c for the benchmark design and the
different optimum secondary configurations for Tf =
290 °C, which corresponds to the HTF temperature at
the inlet of the solar field. At this temperature the highest
temperature gradients occur around the absorber tube.
 M. Wirz et al. / Solar Energy 107 (2014) 398–414 411

potential improvements in thermal efficiency due to ideal

Fig. 15. Absorber tube temperature for different PTC configurations as a
function of the receiver circumferential angle at Tf = 290 °C, which
corresponds to the HTF temperature at the inlet of the solar field. The
asymmetry of the peaks stems from the induced tracking error of the
concentrator mirror.
The angle c = 180° corresponds to the location at the bot-
tom of the absorber tube, facing towards the primary con-
centrator mirror. It can be seen that the benchmark design
exhibits the largest temperature difference of DTabs = 22.7 K
between the coldest and hottest part of the absorber. By
contrast, the seagull design presents a much more balanced
distribution with DTabs = 8.4 K. The incident solar radiation
is absorbed directly or via the involute or wing part of the
secondary, causing several peaks of higher solar flux and
temperature. The maximum temperature difference on the
absorber tube of the aplanatic mirror configuration is also
among the lowest with DTabs = 9.2 K, but the secondary
mirror causes absorption peaks near the top of the absorber,
where little radiation is absorbed. As a result, the tempera-
ture extrema are very close to each other, leading to higher
temperature gradients than in the benchmark design.
selective coating, glass envelope, and primary mirror are
all close to 7%. The potential improvement due to an ide-
alized supporting structure is only about 4%. Table 8 sum-
marizes the results of each individual idealization, together
with the results of a configuration in which all idealizations
are combined. The idealized design has a thermal efficiency
of 81.8%, an improvement of 23% compared to the bench-
mark design. This thermal efficiency comes close to the the-
oretical limit of 86.5% at zero heat losses, where essentially
only cosine losses caused by the solar incidence angle
remain.
To demonstrate the robustness of the results obtained in
the idealization and optimization study, simulations identi-
cal to those described in Section 3.2 were performed for an
alternative PTC design. This alternative design is based on
a concentrator mirror with an aperture width of wconc = 5 m
and a rim angle of wconc = 70°, approximating the Luz LS-2
collector (Dudley et al., 1994). Instead of molten salt, syn-
thetic oil is used as the HTF (DOW, 1997). The number
of concentrator modules per loop is reduced to 24 to limit
the maximum temperature at the outlet of the solar field
to 390 °C. The reflectance curve “Old Cermet ave” from
Kennedy and Price (2005) is used to calculate the spectral
emissivity of a representative previous-generation cermet
selective coating. The remaining design parameters are
identical to those of the benchmark design described in
Section 3.1, including the absorber tube diameter of
Dabs = 0.06 m. Fig. 16 presents the thermal efficiency as a
function of Dabs and wconc for the alternative PTC design.
For consistency, the limits of wconc were selected to be iden-
tical to those in Fig. 1, so the actual alternative design with
wconc = 70° is not included in the figure. Comparison with

4. Discussion
In Section 3.2, potential improvements in the perfor-
mance of PTC systems were systematically evaluated. The

Table 8
Simulation results for configurations with a single idealization and an ideal
design with all idealization combined, together with the increases in
thermal efficiency compared to the benchmark design.
Configuration gopt (%) q_ 0loss (W/m) gth (%) Dgth (%)
Benchmark design 65.0 229 58.8 –
Ideal absorber 68.1 82 65.8 7.0
Ideal glass 68.2 75 66.1 7.3
Fig. 16. Contour plot of gth as a function of Dabs and wconc for an
Ideal mirror 71.9 228 65.6 6.8
alternative PTC design, assuming a concentrator mirror with wconc = 5 m,
Ideal structure 69.0 218 63.1 4.3
a synthetic oil HTF with a maximum temperature of 390 °C, and a
Ideal design 83.8 74 81.8 23.0
previous-generation cermet selective coating.
412 M. Wirz et al. / Solar Energy 107 (2014) 398–414

Table 9 efficiency is reached for a rim angle of 110° coupled to a

Simulation results for the alternative PTC design, assuming wconc = 5 m, slightly smaller absorber diameter than that used in con-
wconc = 70°, a synthetic oil HTF with a maximum temperature of 390 °C,
and a previous-generation selective coating. Results shown for the initial
ventional systems at the expense of a heavier structure
design, individual idealizations, and all idealizations combined. and additional material costs. Idealizing the optical proper-
Configuration gopt (%) q_ 0loss (W/m) gth (%) Dgth (%)
ties of the selective coating, the glass envelope, and the
concentrator mirror up to the theoretical limit result in
Alternative design 64.9 216 57.7 –
Ideal absorber 68.6 29 67.5 9.8
improvements in the thermal efficiency of about 6.8–
Ideal glass 68.1 34 66.8 9.1 7.3%, while a system without shading from the support
Ideal mirror 71.3 216 64.0 6.3 structure gives an increase of around 4.3%. Combining
Ideal structure 69.0 208 62.1 4.4 all idealized components would allow increases of up to
Ideal design 84.4 28 83.2 25.5 23.0% in the thermal efficiency. Future improvements can
be compared with this theoretical limit. These conclusions
are robust in the sense that similar results were obtained for
Fig. 1 shows that both the overall shape of the contour lines
an older-generation, lower-temperature design in addition
and the values of gth are similar to those of the typical PTC
to the current-generation, high-temperature design.
design discussed in Section 3.1. This can be explained by the
The use of several secondary mirror designs was also
relative importance of competing effects. On the one hand,
investigated. The resulting improvements in thermal effi-
the alternative design exhibits a better optical performance
ciency relative to the benchmark design range from 0.8%
than the typical design due to the larger absorber size rela-
for the tailored seagull to 1.6% for the aplanatic mirror.
tive to the aperture area of the concentrator mirror. On
These gains in thermal efficiency are smaller than those
the other hand, the smaller aperture area reduces the solar
obtained with material improvements, but can be put into
input to the system, thereby reducing the heat gained by
perspective by comparison with the potential increase in
the HTF. Furthermore, compared to the ENEA coating
thermal efficiency of a system with no mirror surface and
used in the previous simulations, the “Old Cermet Ave”
tracking errors, where an improvement of 1.2% was deter-
coating has a higher total emissivity, but this is compensated
mined for Dabs = 0.06 m (see Table 5). It should be noted
by the lower overall temperatures of the alternative design,
that while the best performance was achieved using apla-
leading to slightly smaller heat losses compared to the
natic mirrors, even very simple secondary mirror designs
typical design. (It should be noted that this study analyzes
such as the reflective glass surface and the reflective
only the performance of the solar field. If the efficiency of
annulus insulation deliver increases in thermal efficiency
a subsequent power block would be taken into account,
of 1.0% and 0.9%, respectively.
the overall solar-to-electric efficiency of a molten-salt PTC
Most of the improvements presented here rely on ideal
system with temperatures around 600 °C would be signifi-
behavior and materials that may be difficult or impossible
cantly higher than that of an oil-based design at 400 °C).
to achieve in practice. Some of the secondary-mirror designs
The results for the alternative design with wconc = 70°
also present challenges concerning practicality, manufactu-
and Dabs = 0.06 m are compiled in Table 9. The table indi-
rability, and cost. In addition, the benefit of adding second-
cates that the absolute values differ somewhat from the ide-
ary optics appears questionable, as factors influencing the
alization study of the typical system summarized in
optical performance may outweigh better thermal perfor-
Table 8. The lower overall temperatures of the alternative
mance. Nevertheless, the results presented here point to
design lead to lower heat losses, giving higher thermal effi-
the most promising directions for development to increase
ciencies and larger improvements compared to the typical
the thermal efficiency of PTC systems. To determine the
PTC system. Nevertheless, the same overall trends are
most cost-effective approach, dedicated structural, manufac-
observed: The improvement potentials of the selective coat-
turing, and economic analyses must be carried out. Such
ing and the glass envelope are similar and larger than the
analyses are beyond the scope of this work.
improvement potential of the concentrator mirror, and
the potentials of these three components are significantly
larger than the improvement potential of the supporting Acknowledgments
structure, which is around 4% in both designs.
Funding by the European Union under the 7th Frame-
work Program, Grant Nr. 256830 (Project HITECO; New
5. Conclusions solar collector concept for high temperature operation in
CSP applications) is gratefully acknowledged.
The optimum concentrator rim angle and absorber tube
diameters with respect to the thermal efficiency of the solar References
field were determined for a typical current-generation,
Al-Ansary, H., Zeitoun, O., 2013. Heat loss experiments on a non-
high-temperature PTC design. The geometric optimization
evacuated parabolic trough receiver employing a thermally insulating
revealed the improvement potential of the PTC system layer in the annular gap. In: Proceedings of the ASME 2013 7th
without implementing additional components or new International Conference on Energy Sustainability. ASME, Minneap-
materials. The results indicate that a maximum thermal olis, MN.
 M. Wirz et al. / Solar Energy 107 (2014) 398–414
Archimede Solar Energy, 2013. HCEMS11 – Technical Data, <http://
www.archimedesolarenergy.com/hcems11.pdf> (Accessed 10/2013).
Bakos, G.C., Ioannidis, I., Tsagas, N.F., Seftelis, I., 2001. Design,
optimisation and conversion-efficiency determination of a line-focus
parabolic-trough solar-collector (PTC). Appl. Energy 68, 43–50.
Bendt, P., Rabl, A., Gaul, H.W., Reed, K.A., 1979. Optical analysis and
optimization of line focus solar collectors. SERI/TR-34-092. Solar
Energy Research Institute, Golden CO.
Benitez, P., Minano, J.C., Garcia, R., Arroyo, R.M., 1997. Contactless
two-stage solar concentrators for tubular absorber. In: Proceedings of
the SPIE 3139, Nonimaging Optics: Maximum Efficiency Light
Transfer IV. SPIE, San Diego, CA.
Brunotte, M., Goetzberger, A., Blieske, U., 1996. Two-stage concentrator
permitting concentration factors up to 300 with one-axis tracking.
Sol. Energy 56, 285–300.
Buie, D., Monger, A.G., Dey, C.J., 2003. Sunshape distributions for
terrestrial solar simulations. Sol. Energy 74, 113–122.
Burkholder, F., Kutscher, C., 2009. Heat Loss Testing of Schott’s 2008
PTR70 Parabolic Trough Receiver. Technical Report, NREL/TP-550-
45633, National Renewable Energy Laboratory, Golden CO.
Canavarro, D., Chaves, J., Collares-Pereira, M., 2013. New second-
stage concentrators (XX SMS) for parabolic primaries; comparison
with conventional parabolic trough concentrators. Sol. Energy 92,
98–105.
Chaves, J., 2008. Introduction to Nonimaging Optics, first ed. CRC Press.
Chaves, J., Collares-Pereira, M., 2000. Ultra flat ideal concentrators of
high concentration. Sol. Energy 69, 269–281.
Coastal Chemical Company, LLC, 2004. HITECÒ Heat Transfer Salt.
<http://www.coal2nuclear.com/MSR%20-
%20HITEC%20Heat%20Transfer%20Salt.pdf> (accessed 10/2013).
Collares-Pereira, M., 1979. High temperature solar collector with optimal
concentration: non-focusing Fresnel lens with secondary concentrator.
Sol. Energy 23, 409–420.
Collares-Pereira, M., Gordon, J.M., Rabl, A., Winston, R., 1991.
High concentration two-stage optics for parabolic trough solar
collectors with tubular absorber and large rim angle. Sol. Energy
47, 457–466.
Cooper, T., Steinfeld, A., 2011. Derivation of the angular dispersion error
distribution of mirror surfaces for monte carlo ray-tracing applica-
tions. J. Sol. Energy Eng. 133, 0445011–0445014.
DOW, 1997. Syltherm 800 Heat Transfer Fluid, Product Technical Data.
The Dow Chemical Company, Midland, MI.
Dudley, V.E., Kolb, G.J., Mahoney, A.R., Mancini, T.R., Matthews,
C.W., Sloan, M., Kearney, D., 1994. Test Results: SEGS LS-2 Solar
Collector. SAND94-1884. Sandia National Laboratories, Albuquer-
que, NM.
Esposito, S., Antonaia, A., Addonizio, M.L., Aprea, S., 2009. Fabrication
and optimisation of highly efficient cermet-based spectrally selective
coatings for high operating temperature. Thin Solid Films 517, 6000–
6006.
Fernández-Garcı́a, A., Zarza, E., Valenzuela, L., Pérez, M., 2010.
Parabolic-trough solar collectors and their applications. Renew.
Sustain. Energy Rev. 14, 1695–1721.
Forristall, R., 2003. Heat Transfer Analysis and Modeling of a Parabolic
Trough Solar Receiver Implemented in Engineering Equation Solver.
NREL/TP-550-34169. National Renewable Energy Laboratory,
Golden CO.
Friedman, R.P., Gordon, J., Ries, H., 1993. New high-flux two-stage
optical designs for parabolic solar concentrators. Sol. Energy 51, 317–
325.
Gee, R., Cohen, G., Winston, R., 2002. A Nonimaging Receiver for
Parabolic Trough Concentrating Collectors. In: Proceedings of the
ASME Solar 2002: International Solar Energy Conference (SED2002).
ASME, Reno, NV.
Goldstein, A., Feuermann, D., Conley, G.D., Gordon, J.M., 2011. Nested
aplanatic optics. In: Proceedings of the SPIE 8124, Nonimaging
Optics: Efficient Design for Illumination and Solar Concentration
VIII. SPIE, San Diego, CA.
413
Gordon, J., Ries, H., 1993. Tailored edge-ray concentrators as ideal
second stages for Fresnel reflectors. Appl. Opt. 32, 2243–2251.
Gordon, J.M., Feuermann, D., 2005. Optical performance at the
thermodynamic limit with tailored imaging designs. Appl. Opt. 44,
2327–2331.
Gordon, J.M., Feuermann, D., Young, P., 2008. Unfolded aplanats for
high-concentration photovoltaics. Opt. Lett. 33, 1114–1116.
Incropera, F.P., DeWitt, D.P., Bergman, T.L., Lavine, A.S., 2006.
Fundamentals of Heat and Mass Transfer, sixth ed. John Wiley &
Sons, New York, NY.
Jenkins, D., Winston, R., 1996. Integral design method for nonimaging
concentrators. JOSA A 13, 2106–2116.
Kalogirou, S.A., Lloyd, S., Ward, J., Eleftheriou, P., 1994. Design and
performance characteristics of a parabolic-trough solar-collector
system. Appl. Energy 47, 341–354.
Kennedy, C.E., Price, H., 2005. Progress in development of high-
temperature solar-selective coating. In: Proceedings of the ASME
2005 International Solar Energy Conference (ISEC 2005), pp. 749–755.
Lienhard, J.H.I., Lienhard, J.H.V., 2011. A Heat Transfer Textbook,
fourth ed. Phlogiston Press, Cambridge, MA.
Lynden-Bell, D., 2002. Exact optics: a unification of optical telescope
design. Mon. Not. R. Astron. Soc. 334, 787–796.
Maccari, A., 2006. ENEA Activities on CSP Technologies. Proceedings of
the Parabolic Trough Technology Workshop. NREL, Incline Village,
NV.
Manzolini, G., Giostri, A., Saccilotto, C., Silva, P., Macchi, E., 2011.
Development of an innovative code for the design of thermodynamic
solar power plants part A: code description and test case. Renewable
Energy 36, 1993–2003.
McIntire, W.R., 1980. Secondary concentration for linear focusing
systems: a novel approach. Appl. Opt. 19, 3036–3037.
Mills, D., 1995. Two-stage solar collectors approaching maximal concen-
tration. Sol. Energy 54, 41–47.
Montes, M.J., Abanades, A., Martinez-Val, J.M., 2010. Thermofluidy-
namic model and comparative analysis of parabolic trough collectors
using oil, water/steam, or molten salt as heat transfer fluids. J. Solar
Energy Eng. 132.
O’Gallagher, J., Winston, R., 1986. Test of a “trumpet” secondary
concentrator with a paraboloidal dish primary. Sol. Energy 36, 37–44.
O’Gallagher, J., Winston, R., 1988. Performance model for two-stage
optical concentrators for solar thermal applications. Sol. Energy 41,
319–325.
Omer, S.A., Infield, D.G., 2000. Design and thermal analysis of a two
stage solar concentrator for combined heat and thermoelectric power
generation. Energy Convers. Manage. 41, 737–756.
Ostroumov, N., Gordon, J.M., Feuermann, D., 2009. Panorama of dual-
mirror aplanats for maximum concentration. Appl. Opt. 48, 4926–
4931.
Patnode, A.M., 2006. Simulation and Performance Evaluation of Para-
bolic Trough Solar Power Plants. Master’s Thesis, University of
Wisconsin-Madison.
Petrasch, J., 2010. A Free and Open Source Monte Carlo Ray Tracing
Program for Concentrating Solar Energy Research. In: Proceedings of
the ASME 2010 4th International Conference on Energy Sustainability
(ES 2010), pp. 125–132.
Price, H., Lupfert, E., Kearney, D., Zarza, E., Cohen, G., Gee, R.,
Mahoney, R., 2002. Advances in parabolic trough solar power
technology. J. Sol. Energy Eng. 124, 109–125.
Ries, H., Spirkl, W., 1996. Nonimaging secondary concentrators for large
rim angle parabolic troughs with tubular absorbers. Appl. Opt. 35,
2242–2245.
Riffelmann, K.J., Neumann, A., Ulmer, S., 2006. Performance enhance-
ment of parabolic trough collectors by solar flux measurement in the
focal region. Sol. Energy 80, 1303–1313.
Schiricke, B., Pitz-Paal, R., Lüpfert, E., Pottler, K., Pfänder, M.,
Riffelmann, K.J., Neumann, A., 2009. Experimental verification of
optical modeling of parabolic trough collectors by flux measurement.
J. Sol. Energy Eng. 131, 0110041–0110046.
414 M. Wirz et al. / Solar Energy 107 (2014) 398–414
Serghides, T.K., 1984. Estimate friction factor accurately. Chem. Eng. 91,
63–64.
Spirkl, W., Ries, H., Muschaweck, J., Timinger, A., 1997. Optimized
compact secondary reflectors for parabolic troughs with tubular
absorbers. Sol. Energy 61, 153–158.
Winston, R., 1980. Cavity enhancement by controlled directional
scattering. Appl. Opt. 19, 195–197.
Winston, R., Gordon, J.M., 2005. Planar concentrators near the étendue
limit. Opt. Lett. 30, 2617–2619.
Winston, R., Miñano, J.C., Benitez, P.G., 2005. Nonimaging Optics, first
ed. Elsevier Academic Press, Amsterdam.
Wirz, M., Roesle, M., Steinfeld, A., 2012. Three-dimensional optical and
thermal numerical model of solar tubular receivers in parabolic trough
concentrators. J. Sol. Energy Eng. 134, 0410121–0410129.
Wirz, M., Roesle, M., Steinfeld, A., 2014. Design point for predicting
year-round performance of solar parabolic trough concentrator
systems. J. Sol. Energy Eng. 136, 0210191–0210197.