Renewable Energy 63 (2014) 263e271

Contents lists available at ScienceDirect

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

Technical note

Mirrors array for a solar furnace: Optical analysis and simulation

results
D. Jafrancesco a, P. Sansoni a, *, F. Francini a, G. Contento b, C. Cancro c, C. Privato c, G. Graditi c,
D. Ferruzzi d, L. Mercatelli a, E. Sani a, D. Fontani a
a

CNR-INO National Institute of Optics, Largo E. Fermi, 6, Firenze 50125, Italy

ENEA Brindisi Research Centre, SS 7 Appia Km 712, Italy
ENEA Portici Research Centre, Via Vecchio Macello, 80055 Portici, NA, Italy
d
INAF, Astrophysical Observatory of Arcetri, Largo E. Fermi, 5, Firenze 50125, Italy
b
c

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 27 November 2012
Accepted 4 September 2013
Available online 10 October 2013

The optical design of a concentration system for a solar furnace is studied, proposing several possible
solutions. The foreseen use of this solar furnace is to test components and methodologies for solar applications. The analysis assesses and compares the optical performances of several possible congurations. The possibility of employing in a solar furnace an array of off-axis mirrors as primary optics is
examined comparing simulations with various diameters and different congurations. In particular the
paper compares spherical mirrors, parabolic mirrors with axis inclined with respect to the heliostat rays
and a paraboloid with axis parallel to the rays arriving from the heliostat. It proposes an optimal solution,
with spherical mirrors on a spherical envelope, which is compared to the heliostat-axis paraboloid.
Considering realisation tolerances, mirrors positioning, mirrors pointing and solar divergence effects
they equivalently concentrate the sunlight on the receiver.
2013 Elsevier Ltd. All rights reserved.

Keywords:
Solar energy
Lighting simulation
Optical design
Concentrating solar power
Solar furnace

1. Introduction
In the last fty years several solar furnaces have been studied,
designed and realised, both for research applications and for materials test [1e7]. A solar furnace is a structure that uses a
concentrated solar beam to produce high temperatures, usually for
sample test. These plants require a captation area inferior to that
used in electricity production plants, but the mirrors system should
generate a higher power density to improve usage exibility. The
optical layout of this plant type is characterised by a heliostat
(composed of at or curve mirrors), a possible primary optics to
concentrate the light (if not focused by the heliostat) and a solar
receiver (sometimes with an extra concentrator before the receiver
entrance). The mirrors eld is realised by means of one or more
matrices of tens or hundreds elements, placed on the ground and
orienting the reected ux towards a receiver on a tower [8]. Heliostat elds in central tower power applications are huge primarily
because high amounts of radiative energy have to be collected; vast
* Corresponding author. Tel.: 39 055 23081; fax: 39 055 2337755.
E-mail addresses: paola.sansoni@ino.it, paola.sansoni@inoa.it (P. Sansoni),
gaetano.contento@enea.it (G. Contento), carmine.cancro@enea.it (C. Cancro),
ferruzzi@arcetri.astro.it (D. Ferruzzi).
0960-1481/$ e see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.renene.2013.09.006

plants show a spot enlargement due to the solar divergence but the
large plant size reduces mutual mirrors shadowing effects [9,10].
The receiver often is located at several tens of meters, complicating
the access and thus needing further structures. The use of few large
size heliostats with xed parabolic secondary is possible: it was
realised for example at Odeillo, in France [11], but it needs huge
funds and a specic soil preparation. Please note that, from an
optical point of view, the best way to concentrate collimated rays
from a source at innite distance (the sun) is to utilise a parabolic
mirror (or a part of it) with the axis parallel to the rays direction
(parallel-axis paraboloid), so to eliminate the spherical aberration
and minimise coma and astigmatism. It is very difcult to practically produce this system, but in Section 3 the parallel-axis paraboloid will represent a reference (reference solution) in order to
evaluate the behaviour of the nally chosen solution.
Due to the difculties to realise a very large single mirror or few
large mirrors as in the Odeillo plant, the standard solution for the
concentration stage is an array of mirrors (facets) of small sizes
with respect to the total system [12e14]. Several studies examined
how to realise and mount the single mirrors [15]: often they foresee
to use as facets various sets of spherical mirrors in a parallel-axis
system conguration (with the axis of the theoretic surface sustaining the mirrors parallel to the solar rays direction) [7,16,17].

264

D. Jafrancesco et al. / Renewable Energy 63 (2014) 263e271

N
flat heliostat

primary mirror

secondary (CPC)

The heliostat is a composite at mirror, with sizes 10 m  12 m

(H  L), and the primary optics must be centred at a distance
27.413 m from the heliostat centre. It is an off-axis concentrator
because there is an angular misalignment of 17.6 between the
normal of the primary centre and the primary-heliostat connection: in particular in the horizontal plane this angle is 16.5 , while
in the vertical plane is 7.5 .
The Compound Parabolic Concentrator, employed as secondary
optics, is located at 15.512 m from the primary optics centre. The
furnace entrance aperture (the CPC output window) should measure 230 mm in diameter, so the CPC input window, for the
maximum input beam inclination (30.3 ), will have a diameter of
603.6 mm. The fundamental request is to obtain at least 30 kW of
maximum ux at the receiver entrance. Due to the furnace scope,
there are no requirements about specic proles of irradiation at
the furnace entrance.
2. Shape and performances of primary optics mirrors

furnace

Fig. 1. Scheme of the ENEA plant (lengths in mm); in the gure, the at heliostat is
oriented at noon, then the rays that hit the heliostat come from South (in the same
vertical planes of rays from the heliostat to the primary mirror).

Indeed for the DLR furnace at Cologne (Germany) another solution

was chosen: a spherical mirrors array mounted on a at layout
inclined with respect to the direction of rays from the heliostat
(inclined-axis setup) [18]. In this case and in the paper of D.
Riveros-Rosas et al. [7], the mirrors have three different focal
lengths to increase the sun concentration. However, in order to
contain the costs the task should be to apply a set of identical
mirrors, determining the best theoretical surface sustaining the
mirrors set. With this aim, power distribution on the target and
beam quality (Seidel aberrations) were analysed, comparing
various congurations.
The furnace is under realisation near Naples (Italy), in a site of
the ENEA-Portici research centre, with 40 490 110064 N of latitude, at
sea level. Fig. 1 summarises the installation scheme.

2.1. Research methodology

A preliminary evaluation, at least rough, of the beam angular
distribution is necessary. It essentially depends on solar divergence,
heliostat pointing errors and mirrors surface features (slope errors
and diffused reectance) [19]. The primary pointing errors are
neglected, assuming that during the system erection phase every
spherical mirror gets actually aligned correctly to point towards the
target centre. The considered contributions have various angular
dependences: the solar divergence has its peculiar form, which for
the purposes can be approximated as a uniform distribution with
semi-aperture 4.7 mrad [20]; the heliostat pointing errors are
almost uniform in time, with a semi-angle depending on the
tracking system characteristics, of about 3 mrad [19]. The assessment of mirrors local slope errors is more complex: supposing a
Gaussian distribution, with standard deviation s 5 mrad [21], and
converting it [22] into a uniform distribution with semi-aperture
a O3$s, it can be obtained an acceptable good approximation of
the real behaviour [23].
Hence the study includes two simulations with different angular
proles:
- uniform distribution with semi-aperture 4.7 mrad (approximate
solar distribution);
- Gaussian distribution with s 5.9 mrad (reporting uniform
distributions to Gaussian and considering that the convolution
variance is the variances sum).

Fig. 2. Heliostat (green line), mirror envelopes (blue spherical, purple parabolic),
target (CPC, light blue) and rays from the sun (direction of rays corresponding to Jun,
21 at noon, Portici, Naples, Italy) reected from heliostat and from the parabolic envelope, that corresponds to the reference solution. The simulated sun is not visible.

It is evidently impossible to construct a single mirror as primary

optics, or a multi-mirror parabolic surface, for costs reasons. So the
simplest solution is to realise a set of identical mirrors that allow to
obtain an acceptable power concentration at furnace entrance.
Hence the task is to design and place an identical mirrors set, arranged to intercept the radiation arriving from the heliostat and to
redirect it towards the furnace with the maximum possible
efciency.
Concerning shape and size of the mirrors, literature [15] and
practical estimations suggested to use hexagonal mirrors with 1 m
of vertexevertex distance. For validation, the simulations considered mirror diameters between 0.5 m and 2 m. Due to the severe
environmental conditions (seaside site), rear-face aluminated
mirrors were preferred, because their glass absorption losses are
acceptable (under 0.1e0.5% between 400 and 1500 nm of wavelength for few mm thickness), while the reection on the mirror
rst surface is anyway directed to the target, without appreciable
differences from the principal reection (both for the reasonable

D. Jafrancesco et al. / Renewable Energy 63 (2014) 263e271

requested concentration and for the long mirrors focal lengths).

This consideration holds also for the at heliostat, which is then
composed of rear-face aluminated mirrors (at mirrors).
More complex decisions concern single mirror prole and mirrors set envelope, which is the surface supporting the mirrors set.
The solution should allow:
- to have a suitable beam concentration at receiver entrance;
- the use of identical mirrors;
- to be easily realisable and controllable.
For the set envelope curve, the constraints impose that the
quadric envelope passes for the primary centre. The simplest solutions are: (1) parabolic envelope (with axis parallel to the beam
axis at heliostat exit and passing for target entrance), (2) plane
envelope and (3) spherical envelope. The plane envelope is discarded since it has neither optical advantages (with off-axis mirrors
the eld curvature is elevated) nor constructive advantages, having
to completely realise the mirrors set support. If the envelope coincides with a parabolic mirror portion working on axis (optically
the best concentrator for rays arriving from innite) good results
could be obtained; this solution will be compared with the one
obtained posing the mirrors on a suitable spherical envelope. This
will be the core of these analyses on the envelope supporting the
mirrors set.
Regarding the mirrors prole, in the examined case the use of
at mirrors would imply a spot minimum dimension, certainly not
inferior to the single mirror size, to which solar divergence effects
and other beam diverging effects could be added. As already estimated [7], this solution is inconvenient. Consequently the promising solutions are substantially two:
- parabolic mirrors;
- spherical mirrors.
Considering primary-furnace distance and solar divergence
presence, the spot on the target will have at least a diameter of
146 mm. It must be noted that the beam has an angular divergence
greater than the solar one and this determines an inevitable further
beam enlargement (due to mirror surface imperfections) that attenuates the differences among the spots generated from the
various proles on the concentrator entrance. For the same reason,
the advantages of possible aspherical proles were judged negligible, while they improved complexity and costs.
Simulations exploited both sequential and non-sequential
methods. They permit to evaluate the behaviour of an optical system in different ways. In a sequential simulation the rays hit all the
surfaces (lens, mirrors) in a well-dened order, set by the user.
Sequential simulations have many advantages: ray tracing speed,
generality of surface shapes and properties, simple optimisation
and tolerancing, capability to evaluate optical aberrations. On the
contrary, in a non-sequential simulation the rays propagate along
the surfaces without a well-dened order (more similarly to the
effective rays propagation). Non-sequential simulation allows more
detailed analysis of light propagation, including scattered or
partially reected light. Selections of single mirror shape and
diameter were guided both by the spot aberrations and encircled
energy analysis, obtained from sequential optical simulations on
single mirrors, while non-sequential simulations will permit to
obtain the spot proles for envelopes both spherical and parabolic.
Finally the selected solution was completely simulated, with the
whole mirror set, using a non-sequential specialised software
(Lambda Research TracePro) to obtain the target irradiation map,
which will be compared with the same map obtained using a
perfect paraboloid replacing the mirror set. The paraboloid axis is

265

Table 1
Transverse aberrations (in mm) in the Gaussian eld; TSPH tangential spherical,
TTCO tangential coma, TAST tangential astigmatism, TPFC tangential Petzval
eld curvature.
Surface

Diameter

TSPH

TTCO

TAST

TPFC

Spherical
Parabolic
Spherical
Parabolic
Spherical
Parabolic
Spherical
Parabolic

0.5

0.01
0
0.06
3.83
0.22
0
0.52
0

0.96
0.96
3.83
50.31
8.63
8.63
15.34
15.34

25.16
25.16
50.31
25.17
75.47
75.47
100.63
100.63

12.58
12.58
25.16

1
1.5
2

37.74
37.74
50.31
50.31

parallel to the rays arriving from the heliostat and crosses the
target; the paraboloid has its focus on the target. This was
considered as a reference solution for the primary mirror because
it is the best optical way to concentrate rays from innite. Obviously
it is very difcult and expensive to realise a parabolical surface of
diameter 8.5 m as a primary mirror, but it is useful as reference to
evaluate the performance of the system that will be chosen.
In Fig. 2, where the Z-axis is the North-South direction and the
X-axis is parallel to the ground, the two possible envelopes are
highlighted: the heliostat (green), the spherical envelope (blue),
the parabolic one (purple), the target (the CPC, light blue). Some
solar rays (with Gaussian distribution, at noon of the June, 21) reected by the heliostat and from the parabolic envelope are also
shown in red. The two envelopes intersect at the origin of the
reference system, that is the centre of the primary mirror. The
spherical envelope is a sphere centred on the target, while the
parabolic envelope, that coincides with the reference solution,
has its focus on the target.
2.2. Sequential simulations
To decide the mirror shape, spherical and parabolic mirrors of
diameters 0.5 m, 1 m, 1.5 m, 2.0 m were compared, assessing the
aberration values (in mm) on the Gaussian (paraxial) focus, obtained by means of the Seidel coefcient. Using off-axis mirrors, the
most relevant third-order aberrations are spherical, coma, astigmatism and eld curvature [7,17,24]. The inevitable contribution
due to spherical aberration introduced on the spherical mirror
(compared to the parabolic mirror) was examined with respect to
the other aberrations. Due to the spherical aberration independence from the eld (in optics, the inclination of the rays with
respect to the optical axis), only the 17.6 eld was considered. The
mirrors curvature radius is 31.024 m. In Table 1 the values of these
aberrations are shown: TSPH is the tangential spherical, TTCO the
tangential coma, TAST the tangential astigmatism, TPFC the
tangential Petzval eld curvature. Please note that the TPFC value is
half and opposite of the TAST value, due to the fact that the sagittal
astigmatism is zero [25].
The data were directly obtained from sequential simulations
carried out by means of Zemax software: the presented parameters
are the output of the simulation software.
From Table 1 it is clear that the spherical aberration is substantially negligible with respect to the other aberrations, for every
mirror diameter. Therefore spherical mirrors are chosen, considering their constructive simplicity, limiting the further studies to
spherical reectors.
In order to determine the best trade-off between performance
and mirrors diameter (that consequently denes the mirror number), in Table 2 the values of aberrations on the best focus (the
medial plane between tangential focus and sagittal focus) are reported. These values are calculated according to Ref. [26], pp. 144e
145; since this procedure will lead to the choice of the envelope for

266

D. Jafrancesco et al. / Renewable Energy 63 (2014) 263e271

Table 2
Transverse tangential coma (in mm), S3 and S4 Seidel coefcients and transverse
astigmatism in medial plane (best focus) for different mirror diameters and elds.
Diameter and
eld

TTCO

S3

S4

Transverse Gaussian
to medial

0.5 m, 8.5
0.5 m, 17.6
0.5 m, 30.3
1 m, 8.5
1 m, 17.6
1 m, 30.3
1.5 m, 8.5
1.5 m, 17.6
1.5 m, 30.3
2 m, 8.5
2 m, 17.6
2 m, 30.3

0.4516
0.9586
1.766
1.806
3.834
7.063
4.065
8.627
15.892
7.226
15.337
28.253

0.089993
0.405443
1.375823
0.359973
1.621772
5.503291
0.809939
3.648986
12.382405
1.439892
6.487087
22.013165

0.089993
0.405443
1.375823
0.359973
1.621772
5.503291
0.809939
3.648986
12.382405
1.439892
6.487087
22.013165

0.697985708
3.144615908
10.67088319
1.395975294
6.289231816
21.3417625
2.093962295
9.433845139
32.01264439
2.791950588
12.57846169
42.68352694

having the receiver in the best focus for all the mirrors (see Section
2.3), the effective values of aberrations, which have to be compared
with the beam enlargement, must be calculated on the medial
plane.
It must be consider that a Gaussian beam enlargement with
s 5.9 mrad brings at 15.5 m to a spot radius at 3*s (where about
the 99% of the energy is concentrated) of 274 mm. The situation
changes if only the solar enlargement of the beam is considered; in
fact, the spot radius in this case is about 73 mm; it represents an
inferior limit for the performance of the system. In order to not
decrease the solar concentration, the spot radius due to aberrations
has to be signicantly less than these values for all the elds, then
the diameter of 2 m is not acceptable. To assess the better trade-off
it is also possible to consider the encircled energy radii (for 99% of
the total energy of the spot) on the best focus surface for different
spherical mirror diameters and elds, as shown in Table 3. These
values, another direct output of the sequential simulations, show
that if the mirror diameter of 1 m is chosen, the encircled energy
radii of the largest eld (30.3 ) are 79.9 mm, a value comparable
with the 73 mm of the solar enlargement of the mirror.
Then the diameter of 1 m is the better trade-off between costs
and performances: in fact, a decrease of the mirror diameter does
not bring to a signicantly better spot due to solar enlargement.
The literature conrms that a mirror frame with facets of
diameter 1 m permits to concentrate the energy in a similar way
with respect to a frame mirror with facets of minor dimensions
when the ratio between frame focal distance and frame diameter is
larger than 1.2 [15]; in the conguration of Fig. 2 this ratio is 1.8, but
the incoming beam is not parallel to the frame axis.
2.3. Non-sequential simulations
2.3.1. Mirror envelope supporting the mirrors
For the envelope curve of the mirror set, the constraints are that
the quadric envelope intersects the primary centre. The rst
reasonable hypothesis is to accurately approximate, using spherical
mirrors, the paraboloid with axis parallel to the axis of the beam at
heliostat exit and crossing the target entrance. In this way the
mirrors set would coincide with a parabolic mirror portion working
Table 3
Encircled energy radii (for 99% of the total energy of the spot).

Mirror
Mirror
Mirror
Mirror

diam
diam
diam
diam

0.5 m
1m
1.5 m
2m

Field 8.5

Field 17.6

Field 30.3

2.9
6.3
10.4
15.2

12.3
25.9
40.4
56.8

39.1
79.9
125.3
173.2

on-axis ([15] preferred this conguration). However, Table 1 indicates that the highest aberrations contributions come from
tangential astigmatism and eld curvature that quadratically
depend from the eld ([27], pp. 210e211). In fact, the elevated
astigmatism introduced by these congurations was already been
noticed [24]. To compensate, at least partially, this effect it can be
supposed to place the mirrors on a spherical envelope surface that
has the same radius of the Petzval surface of the mirrors, which is a
sphere with radius RPetz 15.512 m. This allow the best focus
(medial focus between tangential and sagittal focus) of each mirror
to coincide with the receiver centre, simplifying the check operations on the set envelope surface. A further advantage is that the
spot on the receiver, without other aberrations, would have a circular form (useful for irradiation uniformity on the secondary or for
uses without secondary optics).
Using non-sequential Zemax simulations the beam enlargement
(Gaussian or solar) can be taken into account; the behaviour of the
two congurations, parabolic and spherical envelope, can be
controlled in signicant set points, separately assessing the spot
size for the spherical envelope and the parabolic envelope.
Thus, all the following parameters were combined in nonsequential Zemax simulations:
- spherical mirrors on spherical and parabolic envelope;
- angles between incident rays and normal to the mirrors at 8.5
(the less inclined mirror of the set), at 17.6 (central mirror of the
set) and at 30.3 (extreme mirror of the set);
- two beam divergences (uniform with semi-aperture 4.7 mrad
and
Gaussian
distribution with
standard
deviation
s 5.9 mrad 0.338 ).
The spot transversal dimensions on the target (at the entrance of
the secondary, placed at a distance 15.512 m from primary centre),
at 10% of the maximum power density, have been reported. Obviously, the more the spot size is reduced, the more the conguration
is preferable; moreover, in order to optimise the ux distribution
on the target, it is advantageous to have a circular spot rather than
an oval spot, therefore also the spot astigmatism has to be
considered.
Fig. 3 reports average spot diameter versus angle between rays
and mirror axis, both for uniform enlargement and for Gaussian
enlargement. Fig. 4 presents the spot astigmatism, i.e. the ratio
between spot vertical and horizontal diameters, separating the two
beam enlargements.
For the spot size, Fig. 3 shows that the spherical envelope permits to have the best results both for uniform beam enlargement
and for Gaussian beam enlargement. Referring to the spot astigmatism (Fig. 4), since the best condition is around the value 1, also
in this case the spherical envelope represents an advantageous
choice.
2.4. Final conguration
The conclusive decisions for the design of the heliostat plant
conguration can be summarised as follows:
- utilise spherical mirrors of diameter 1 m and curvature radius
31.024 m;
- dispose these mirrors on a spherical surface centred on the
furnace and crossing the centre of mirrors set.
The nal scheme is illustrated in Fig. 5: the heliostat is shown in
green colour, the mirror set in azure, a secondary optics (CPC,
Compound Parabolic Concentrator) in blue; the rays arriving from
the sun (in red) are reected by the heliostat and the mirror array.

D. Jafrancesco et al. / Renewable Energy 63 (2014) 263e271

267

Fig. 3. Spot size.

The input window of the CPC is the target of the primary mirror set,
and its centre is at 15.512 m away from the primary mirror set. The
mirrors have different curvature radii and the axis of a single mirror
cannot be coincident with a radius of the envelope, so the surface of

a single mirror does not coincide with the envelope surface, but
they are arranged in order to avoid energy losses: in practice, the
total reecting surface is seen as a continuum from the heliostat.
3. Comparison between inclined-axes spherical mirrors set
and parallel-axis unique paraboloid

Fig. 5. The complete system: heliostat in green, primary in azure, CPC in blue; the rays
arriving from the sun (in red) are reected by the heliostat and the mirror array. (For
interpretation of the references to colour in this gure legend, the reader is referred to
the web version of this article.)

It is useful to evaluate the behaviour of the entire spherical

mirrors set [28] with respect to the paraboloid with axis parallel to
the rays arriving from the heliostat (reference solution, parallelaxis paraboloid). Having the purpose of using these simulations
also to dimension the furnace components, the maximum power
that could reasonably enter on the furnace will be simultaneously
calculated.
Since the heliostat-primary axis follows the South-North direction, the value at noon of June 21st will be used as a maximum
solar irradiance; in this date, also if in Summer the atmosphere
transparency is averagely inferior, the Air Mass reaches a minimum
value, so the power sent to the primary should attain a maximum.
The literature data [29] furnish an irradiance value for Air Mass 1
of about 970 W/m2, consequently the total ux incident on the
heliostat that arrives on the paraboloid (considering a circular
source of radius 4.0 m) is 48,757.5 W. If the system would be perfect
excluding the surface reectance, and imposing a 90% reectance,
in the CPC should enter a ux of 39,493.6 W.
Simulations, in Lambda Research TracePro software, were performed using as a target a circular surface with diameter 603.6 mm,

Fig. 4. Spot astigmatism.

268

D. Jafrancesco et al. / Renewable Energy 63 (2014) 263e271

Fig. 6. Set of spherical mirrors with collimated beam.

placed in correspondence of the secondary entering window [28].

The reflectance of both heliostat and primary mirrors was set to
90%. For a comparison, some simulations using the parallel-axis
parabola (with 90% of reectivity) were also executed to replace
the primary with facets mirrors. The results for spherical mirrors
and paraboloid are examined in Figs. 6e11, for various beams.

The spherical mirrors set shows a valuable behaviour with

respect to the reference solution in collimated light (Figs. 6 and 7),
being capable to concentrate nearly all the light inside a 50 mmradius circle, a value of the same order of the Seidel aberrations
calculated at the Section 2.2. Obviously the performances degrade
in non-collimated light: with a uniform divergence of semi-angle

Fig. 7. Parabolic mirror with collimated beam.

D. Jafrancesco et al. / Renewable Energy 63 (2014) 263e271

269

Fig. 8. Set of spherical mirrors with uniformly diverging beam, semi-angle 4.7 mrad.

0.27, equal to the solar divergence, substantially all the radiation

enter in the secondary (Figs. 8 and 9); setting a Gaussian divergence, with s 0.338 , 99.3% of the generated radiation can enter
into the secondary. However it is interesting to note that, with
Gaussian divergence, the facets mirrors system substantially
presents the same performances of a parallel-axis paraboloid
(Figs. 10 and 11).

4. Conclusion
A concentration system for a solar furnace, a structure using
concentrated sunlight as a direct source of heat, was optically
designed and analysed, proposing different suitable solutions.
The optical design studies simulated and compared the optical
performances of selected potential congurations, exploiting tools,

Fig. 9. Parabolic mirror with uniformly diverging beam, semi-angle 4.7 mrad.

270

D. Jafrancesco et al. / Renewable Energy 63 (2014) 263e271

Fig. 10. Set of spherical mirrors with Gaussian beam divergence, s 0.338 .

functions and parameters of both sequential and non-sequential

simulations.
The simulations analysed two different reectors, spherical and
parabolic mirrors, and two set envelopes, spherical and parabolic.
In order to evaluate the performance for different positions of the

mirrors on the set, three optical elds (angle between the optical
axis of a mirror and the incoming light direction) are considered:
for mirrors at the centre and at the opposite sides of the set,
respectively 17.6, 8.5 and 30.3 . Moreover, the non-sequential
simulations were performed with two beam enlargements:

Fig. 11. Parabolic mirror with Gaussian beam divergence, s 0.338 .

D. Jafrancesco et al. / Renewable Energy 63 (2014) 263e271

rectangular of semi-angle 0.27 (similar to the solar enlargement)

and Gaussian with s 0.338 (to take into account manufacturing
and mounting tolerances).
By means of sequential simulations rst the equivalence of
spherical mirrors and parabolic mirrors was shown, then it was
assessed the acceptability of mirror diameter of 1 m. This is due to
the fact that the aberrations are signicantly lower than the beam
enlargement, also if only the solar divergence is taken into account.
Thus, in order to dene the best envelope between parabolic surface and spherical surface (with radius equal to the Petzval radius of
the mirrors), by means of non-sequential Zemax simulations the
radii of the spot with solar or Gaussian divergence were calculated
for various elds and the two envelopes. The data showed that the
spherical layout corresponds to the best results (the spot size is
almost always the smaller both for solar and Gaussian enlargement,
and the spot shape is nearby circular). Therefore, a set of spherical
mirrors of radius 1 m laying on a spherical envelope (centred on the
target and with radius equal to their Petzval radius) appears to be
the best conguration. To validate this choice, by means of a
TracePro non-sequential simulation, the spots generated by a
reference paraboloid (a hypothetical single paraboloid with axis
parallel to the rays arriving from the heliostat) and the chosen
mirror set were compared. They differ only for a collimated beam,
instead they are very similar when a solar or Gaussian enlargement
is considered, conrming the choice of the mirror set.
Hence it can be concluded that a layout with spherical mirrors
on a spherical envelope is the best solution to optically design a
solar furnace suitable to test components and methodologies for
solar applications. When a reasonable beam enlargement is
considered, the spherical mirrors on a spherical envelope system
equivalently concentrate the sunlight on the receiver with respect
to the heliostat-axis paraboloid.

References
[1] Trombe Felix. Solar furnaces and their applications. Sol Energy 1957;1(2e3):
9e15.
[2] Hisada T, Mii H, Noguchi C, Noguchi T, Hukuo N, Mizuno M. Concentration of
the solar radiation in a solar furnace. Sol Energy 1957;1:14e8.
[3] Kevane CJ. Construction and operation of the Arizona State College solar
furnace. Sol Energy 1957;1:99e101.
[4] Trelov VI, Schur DV, Pishuk VK, Zaginaichenko SY, Choba AV, Nagornaya NR.
The solar furnaces for scientic and technological investigation. Renew Energy
1999;16(1e4):757e60.

271

[5] Fernndez-Reche J, Caadas I, Snchez M, Ballestrn J, Yebra L, Monterreal R,

et al. PSA solar furnace: a facility for testing PV cells under concentrated solar
radiation. Sol Energy Mater Sol Cells 2006;90:2480e8.
[6] Gordon JM, Babai D, Feuermann D. A high-irradiance solar furnace for
photovoltaic characterization and nanomaterial synthesis. Sol Energy Mater
Sol Cells 2011;95:951e6.
[7] Riveros-Rosas D, Herrera-Vzquez J, Prez-Rbago CA, Arancibia-Bulnes CA,
Vzquez-Montiel S, Snchez-Gonzlez M, et al. Optical design of a high
radiative ux solar furnace for Mexico. Sol Energy 2010;84(5):792e800.
[8] Gttsche Joachim, Buck Reiner, Badstbner Kathrin, Iand David, Teufel Edgar,
Hoffschmidt Bernhard, et al. Solar concentrating systems using small mirror
arrays. J Sol Energy Eng 2010;132(1):010201.
[9] Riaz MR. A theory of concentrators of solar energy on a central receiver for
electric power generation. ASME J Eng Power 1976;98(3):375e83.
[10] Collado Francisco J. Preliminary design of surrounding heliostat elds. Renew
Energy 2009;34(5):1359e63.
[11] Trombe F, Le Phat Vinh A. Thousand kW solar furnace, built by the National
Center of Scientic Research, in Odeillo (France). Sol Energy 1973;15:57e61.
[12] Paderin LY. Composite paraboloidal radiant energy concentrators. Appl Sol
Energy 1978;14(5):45e51.
[13] Umarov GY, Zakhidov RA, Khodzhaev AS. Concentrating power of paraboloidal facets. Appl Sol Energy 1976;12(4):16e21.
[14] John V, Natarajan G, Mohanan P, Mohandas PN, Raman S, Madhava MR, et al.
A mosaic glass reector solar furnace. Energy 1984;9(5):447e52.
[15] Riveros-Rosas D, Snchez-Gonzlez M, Arancibia-Bulnesc CA, Estrada CA. Inuence of the size of facets on point focus solar concentrators. Renew Energy
2011;36(3):966e70.
[16] Carlson DEE, Diver RB, Fletcher EA. A simple model for predicting the ux
distribution through the focal plane of a multifaceted concentrator solar
furnace. J Sol Energy Eng 1984;106(1):103e5.
[17] Diver RB, Carlson DEE, MacDonald FJ, Fletcher EA. A new high-temperature
solar research furnace. ASME J Sol Energy Eng 1983;105(3):288e93.
[18] Neumann A, Groer U. Experimenting with concentrated sunlight using the
DLR solar furnace. Sol Energy 1996;58(4e6):181e90.
[19] Badescu V. Theoretical derivation of heliostat tracking errors distribution. Sol
Energy 2008;82:1192e7.
[20] Buie D, Monger AG, Dey CJ. Sunshape distributions for terrestrial solar simulations. Sol Energy 2003;74:113e22.
[21] OGallagher J, Winston R. Performance model for two-stage optical concentrators for solar thermal applications. Sol Energy 1988;41(4):319e25.
[22] Ente Nazionale Italiano di Unicazione (UNI). Guida allespressione dellincertezza di misura. UNI CEI ENV 13005:2000; notes of point 4.3.9; July 2000.
[23] Harris JA, Duff WS. Focal plane ux distribution produced by solar concentrating reectors. Sol Energy 1981;27(5):403e11.
[24] Zaibel R, Dagan E, Karni J, Rie Harald. An astigmatic corrected target-aligned
heliostat for high concentration. Sol Energy Mater Sol Cells 1995;37:191e202.
[25] Schroeder DJ. Astronomical optics. Academic Press Inc.; 1987. p. 64e87.
[26] Welford WT. Aberrations of optical systems. Bristol and Boston: Adam Hilger
Ltd; 1986. p. 144e5.
[27] Shannon RR. The art and science of optical design. USA: Cambridge University
Press; 1997.
[28] Sing Lim Chern, Li Li. Flux distribution of solar furnace using non-imaging
focusing heliostat. Sol Energy 2009;83(8):1200e10.
[29] Commission Internationale de lclairage. Solar spectral irradiance. Publ. N
CIE 85. 1st ed.; 1989.