Codes for solar flux calculation dedicated to central

receiver system applications: A comparative review

Pierre Garcia, Alain Ferriere, Jean-Jacques Bézian

To cite this version:

Pierre Garcia, Alain Ferriere, Jean-Jacques Bézian. Codes for solar flux calculation dedicated to
central receiver system applications: A comparative review. Solar Energy, Elsevier, 2008, 82 (3),
pp.189-197. �10.1016/j.solener.2007.08.004�. �hal-01717489�

HAL Id: hal-01717489

https://hal.archives-ouvertes.fr/hal-01717489
Submitted on 20 Jun 2018

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Codes for solar flux calculation dedicated to central receiver
system applications: A comparative review
a,*
Pierre Garcia , Alain Ferriere a, Jean-Jacques Bezian b

a
CNRS-PROMES Laboratory, 7 rue du four solaire, 66120 Font-Romeu, France
b
CNRS-LGPSD
Laboratory,
Campus
Jarlard,
81013
Albi
Cedex
9,
France

Abstract

As
we
need
adapted
software
to
calculate
the
solar
flux
concentration
through
a
tower-heliostat
field
system,
an
overview
of
computer

codes
was
performed,
detailing
their
features,
strengths
and
weaknesses.
For
this
a
questionnaire
was
sent
to
developers
or
heavy
users
of

codes
that
are
currently
used
in
the
concentrated
solar
power
(CSP)
community.
Answers
to
this
questionnaire
make
it
possible
to
deter-
mine
which
tool
is
relevant
depending
on
the
objectives
of
the
simulation.
Modeling
tools
for
central
receiver
systems
(CRS)
can
be

divided
into
two
main
categories,
corresponding
to
two
kinds
of
problems:
on
one
hand
those
dedicated
to
system
optimization
and

on
the
other
hand
those
designed
to
detailed
analysis
of
the
optical
performances.
A
bibliographic
study
on
first
generation
and
devel-
oping
codes
complements
this
overview
of
tools
that
may
be
interesting
for
CSP
research
or
industry.

Keywords: Solar flux calculation; Central receiver systems; Optical modeling; Ray-tracing; Cone optics

1. Introduction comparison had already been initiated in the frame of

SolarPACES (solar power and chemical energy systems,
To optimize and design a CRS it is essential to know the IEA implementing agreement) to validate optical tools on
performances of the subsystem formed by the tower and a standard test case (Pitz-Paal and Schwarzbözl, 2000),
the heliostat field. Experimental work in this field has but to our knowledge this work was not continued.
shown the necessity to master user-friendly modeling tools Concerning heliostat fields, two kinds of problems can
to design, simulate and optimize solar components of be distinguished. The first one is an optimization problem:
tower plants. Unluckily a lot of work is generally needed ‘‘What is the best heliostat field layout to maximize the col-
to adapt the codes for concentrated solar flux calculation lected solar energy or to minimize the cost of that energy?’’;
to specific features and specific needs of each project. None the second one is a performance calculation: ‘‘What is the
of such codes can be considered as a standard tool for power reflected by the field and arriving on the receiver
research or industry applications. Our purpose is not to aperture?’’ The PROMES laboratory is interested in both
develop another tool – many of them are available and problems. Indeed, one goal is to be able to optimize the
effective – but to screen existing codes, trying to determine design of any CRS project at any location on economic cri-
which one should be used depending on the objectives of teria such as cost of the produced electricity. An additional
the simulation, and which results can be expected. Such goal is to assess the performances of the solar field for the
PEGASE (Garcia et al., 2006) project located at THEMIS
*
Corresponding author. Tel.: +33 468 307 714; fax: +33 468 302 940. (Targasonne, France, see Fig. 1). To answer this double
E-mail address: pierre.garcia@promes.cnrs.fr (P. Garcia). objective, the needed tool must be able to determine flux
 Fig. 1. View of the Themis facility and scheme of the Pegase system.

maps on receiver aperture, to establish efficiency matrixes to model transient behavior of CSP systems, during start-
of the solar field, to optimize the field layout on costs cri- up, shut-down or variable weather periods. However this
teria (cost of produced kilowatt-hour, cost of installed kilo- third kind of problem, very useful to define operation strat-
watt), and to calculate instantaneous or annual egies of a CRS, is not tackled in this paper.
performances of the concentrating system.
Furthermore, an optional feature is to include receiver
and thermal cycle models to simulate the whole installation 2. Abbreviations and glossary
and to analyze its performances, as two of the screened
codes do. This global simulation on an annual basis can • Abbreviations: In this document the following defini-
also be done by integrating the heliostat field and tower tions are used:
subsystem in less specific tools, such as TRNSYS
(Schwarzbözl et al., 2002), SOLERGY (Stoddard et al., CRS Central receiver HSGT Hybrid solar and
1987), or ECOSTAR (Pitz-Paal et al., 2005). The simplified system gas turbine
model developed for the ECOSTAR program combines CSP Concentrating solar PT Parabolic trough
steady-state component efficiencies from experimental or thermal power
simulated curves. This method is relevant to assess an DS Dish-stirling PV Photovoltaic
annual production, but not to simulate the operation of a
whole installation on shorter periods when the system • Solar field efficiency: reflected power arriving on the
dynamics should be considered. TRNSYS may be adapted receiver aperture divided by the product of the incident
 solar power by the total area of mirrors. It includes of normal Gaussian distributions corresponding to each
reflectivity of the mirrors, cosine factor due to the inci- error (sun shape and heliostat errors).
dent angle of the sun on the heliostats, atmospheric A systematic comparison shows that with the same
attenuation between the heliostats and the receiver, hypothesis similar results can be reproduced with ray-trac-
shadowing and blocking effects, and spillage of the flux ing and convolution methods (Pitz-Paal and Schwarzbözl,
around the receiver aperture. For the mean annual effi- 2000). Simulation errors often come from an incomplete
ciency, heliostat availability is also taken into account. description of reflective surfaces and sun shape properties.
• Shadowing and blocking effects (S&B): a shadow is a Nevertheless ray-tracing methods are more flexible and are
mask (neighboring heliostat, tower) in the incident ray able to model non-ideal optics (non-imaging concentra-
path between the sun and the heliostat surface; a block tors). Indeed they have the advantage of reproducing real
is a mask (neighboring heliostat) in the reflected ray interactions between photons and therefore of giving accu-
path between the heliostat surface and the receiver. rate results for small or complex systems but they need
• Solar field efficiency matrix: bi-dimensional matrix giv- higher calculating time and computing power. That is
ing the solar field efficiency as a function of sun position why it is not recommended to use ray-tracing techniques
(azimuth and elevation). Integrated in tools for perfor- for system optimization.
mance analysis of CSP systems, such a matrix makes it About the accuracy of codes using convolution/expan-
possible to assess the solar field efficiency all year long. sion techniques (Walzel et al., 1977) found peak flux error
Generally speaking, reflectivity of the mirrors is not and average absolute error in the range of 1–2%. This is
included because it may vary during a year. comparable to what can be obtained from a ray-tracing
codes like SOLTRACE. Kistler (1986) states that annual
performances of a solar field can be deduced with accura-
3. Some principles of heliostat field modeling cies better than 1%. Comparison between first generation
codes on a small field (CESA-1) shows very good agree-
The optical components of a CRS are designed to form ment (about 1% deviation for the power on receiver, 3%
an image of the solar disk on a focal plane. However the for the peak flux). However the accuracy to predict flux
obtained solar spot has neither the same size nor the same and spillages decreases when the slant range decreases with
luminance as the sun, because of the following phenomena respect to heliostat and receiver size and in particular with
(Hénault, 1987): high precision heliostat (low canting, focusing and pointing
errors). Nevertheless one have to keep in mind that the
– sun and collector geometry greatest errors observed on annual performance of a
• size and luminance of the sun, varying every day by CRS do not come from the optical model but from the
diffusion in the atmosphere, other components model (turbine, storage. . .).
• optical aberrations, like heliostat astigmatism, caused
by the design of the reflective surfaces, 4. First generation codes
– specific defects of solar facilities
• microscopic errors of reflective facets, often consid- First codes designed to calculate solar flux concentration
ered as negligible, in a CRS came from preliminary studies carried out in the
• pointing (or tracking) errors, well known, US for Solar One in the late 70s. Among them one may
• curvature and canting errors of facets or modules. quote HELIOS, MIRVAL and DELSOL, which are still
used today. HELIOS was initially developed for analyzing
To calculate concentrated solar irradiance, usual experiments at Sandia’s CRTF (Biggs and Vittitoe, 1977).
approaches are ray-tracing and convolution methods. This code can be adapted to all kind of small-sized plants
The principle of ray-tracing methods (or statistical or and has become a widespread tool for facility comparison
Monte Carlo methods) is to choose randomly a bundle of used by many research teams. It can model not only CRS
rays coming from a surface 1, and then to determine which but has also been modified to analyze the optical perfor-
of them arrive on a surface 2. The irradiance of an elemen- mances of parabolic troughs (DLR version) or even dishes.
tary surface is proportional to the number of impacting This FORTRAN code is based on a cone optics approach:
rays (Matteı̈, 2005). In the case of a concentrator with flux density is produced combining error cone of the
one reflection, this algorithm is used twice, first between reflected ray and sun shape by convolution of independent
the sun and the reflective surface with an energetic distribu- distributions with Fourier transform. Heliostat facets of
tion corresponding to the sun shape, then between the any shape and any curvature are divided into small elemen-
heliostat facet and the receiver with a statistic law for the tary mirrors reflecting the sun image on the receiver aper-
error distribution related to the defects of the facet. Preci- ture. HELIOS is considered by Izygon et al. (1987) to be
sion and calculating time increase with the number of rays the most flexible and sensitive code of its generation
and the complexity of geometry. In convolution methods because of its low calculation time and its accuracy. How-
(or cone optics), reflected rays from elementary mirrors ever this tool needs a detailed description of heliostats
are considered with error cones calculated by convolutions geometry, it is quite complex and hard to use for an inex-
perienced user (Izygon et al., 1987). It is not adapted to 5.2. Two categories of codes
large heliostat fields, it can neither assess annual perfor-
mances, nor optimize heliostat layout, nor use secondary As it can be observed in Table 2, two main categories
optics. That is why we decided not to include HELIOS in can be distinguished.
our comparative study, contrary to MIRVAL and These two categories show many differences because
DELSOL. they do not have the same objectives. MIRVAL, FIAT-
Simultaneously in France numerous codes were devel- LUX and SOLTRACE give a detailed description of the
oped before and during experiments at Themis. These are reflected power from a heliostat field but include neither
accurate and agree well with US codes from the same per- thermodynamic conversion nor heliostat field optimization
iod (Izygon et al., 1987), but they are not dedicated to fore- contrary to UHC, DELSOL and HFLCAL. These latter
cast performances of other CRS facilities (Hénault, 1987). codes contain more approximations in their resolution
Moreover when they were written computing techniques methods, so they can quickly assess the annual perfor-
and hardware highly differ from current ones, and they mances of a large-size heliostat field but give less accurate
have not been updated since the 80s. On a theoretical point results for small solar fields. To sum up, MIRVAL, FIAT-
of view, advanced mathematical models were used, but LUX, SOLTRACE or even HELIOS model the heliostat
none of these tools has a user-friendly interface. The only field and tower subsystem whereas UHC includes the recei-
code that has evolved after the Themis experiment is now ver, and DELSOL, and HFLCAL simulate the whole CRS
called COSAC (Hénault, 2005). This is a ray-tracing calcu- including receiver and power block.
lation tool without approximation, adapted to complex
optical systems (spatial optical devices, 3-D spectrometers)
but not to heliostat field modeling. Actually the user would 5.3. Main features of each code
need to describe heliostats one by one, to calculate sun
position, to add instructions to account for shadowing 5.3.1. University of Houston codes
and blocking. At the time of Themis, flux measurements UHC codes (sometimes called RCELL suite) are a suite
techniques were still lacking precision, so comparisons of interconnected Fortran77 codes, each with a number of
between measures and calculations made illusive any optional operating modes, dealing primarily with the opti-
improvement in accuracy of simulation codes (Izygon cal design of heliostat fields and receivers. Thermal and
et al., 1987). economic algorithms are incorporated to enable optimiza-
tion, performance and design studies of the complete plant.
5. Currently used codes Three of these codes are dedicated to the optical subsystem
(Falcone, 1986):
To compare in an objective and reliable way the features
of tools currently used in research centers working on CSP – NS (cellwise performance) provides interception and
technologies, a questionnaire was sent to developers or flux data, diurnal and annual flux data for fixed designs.
heavy users of six codes (UHC, DELSOL, HFLCAL, The solar field is divided into cells corresponding to
MIRVAL, FIAT LUX and SOLTRACE). This inquiry regions with uniform heliostat density or fixed number
contained 23 questions about physical and computational of heliostats, or single heliostats and performance is cal-
model used, user interface developed, parameters involved, culated for a representative heliostat in each cell (Lipps
and results obtained. and Vant-Hull, 1978),
– RCELL (cellwise optimization) optimizes solar compo-
5.1. Results of the inquiry nents (heliostat spacings in field, field boundaries, tower
and receiver dimensions) on cost/performance criteria
Table 1 sums up the main characteristics of these codes. obtaining interception factors for the optimization from
These tools have features in common. First all of them a very simple model (for initial trials) or for accurate
use exact physical considerations to calculate cosine factor results from NS outputs (‘‘node files’’),
and S&B, except FIATLUX for which this feature is not – IH (individual heliostat layout and performance code) is
implemented yet. Likewise all codes can be run in Windows a detailed layout processor using RCELL data to specify
environment on standard computers with user-friendly gra- each heliostat location and can also compute perfor-
phic interfaces like flux maps and 3-D visualization of the mance for each heliostat or for the whole field.
installation (Fig. 2) except for MIRVAL. The size of the
solar field is not limited. These tools are in the public Input is contained in four modules defining the site and
domain, except FIAT LUX which availability is pending weather, the heliostat design, the receiver design, and the
on CIEMAT decision according to software policy, and type of run (interception data, field optimization, system
HFLCAL. However source codes of SOLTRACE and optimization, heliostat layout or annual performance). By
some additional programs (WINDELSOL interface by generating and using data bases (‘‘cosine, shading, and
CIEMAT or DLR extensions for MIRVAL) are not avail- blocking files’’ and ‘‘node files’’) computing time is saved
able now. (Falcone, 1986). UHC codes have been used to optimize
Table 1
Main features of five codes for concentrated solar flux calculation
Name of code UHC DELSOL HFLCAL MIRVAL FIAT LUX SolTRACE
References Pitman and Vant-Hull, 1989 Kistler, 1986 Kiera, 1986 Leary and Monterreal, 2000 Wendelin,
Laurence et al., 1984 Lipps Hankins, 2003
and Vant-Hull, 1980a,b 1979
Research University of Houston SANDIA GAST project SANDIA CIEMAT-PSA NREL
team
Currently SANDIA, Tietronix CIEMAT, SANDIA DLR DLR, CIEMAT NREL,
used by SANDIA CNRS
Considered CRS-SF CRS CRS CRS CRS CRS-PT-DS-
technologies SF
Development 1974 1978 1986 1978 1999 1999
start
Availability Source and executable version Source and executable Not available Source and ? Executable
version executable version
version
Programming FORTRAN/C++ FORTRAN/Basic FORTRAN FORTRAN MATLAB Delphi5
language
Flux Hermite polynomial Hermite polynomial Simplified convolution Monte Normally Monte Carlo
calculation expansion/convolution expansion/convolution of each heliostat’s flux Carlo ray- distributed random ray-tracing
method tracing value of ‘slope
errorc
Receiver type Flat, cavity or external Flat, cavity or external Flat, cylindrical or Flat, cavity Flat Almost any
cylinder cylinder conical or external receiver
cylinder configuration
Multiple Beam-down but not secondary No Yes Yes No Yes
reflections concentrators
Contribution Yes Yes Yes Yes No No
of each loss
Annual Yes Yes Yes Yes No No
performances
Optimized Heliostats layout and Heliostat boundary, Heliostat layout, tower Heliostat Not available Not
components boundary, tower height, layouta, tower height, height, receiver area layoutb available
receiver geometry receiver size, storage and orientation
capacity
Optimization Energy or cost criteria with Cost criteria with optional Energy or cost criteria Energy Not available Not
criteria and allowable flux/land constraint flux/land constraints criteriab available
constraints
a
Uses UHC defined spacings.
b
With DLR additional program only.
c
For each unitary normal vector to the mirror surface, then calculation of deterministic trajectories for the bundle of reflected rays coming from solar
disk with geometric optics laws.

heliostat fields and to evaluate optical performance of a mance calculations than either MIRVAL or HELIOS (Fal-
number of CRS, including solar one and solar two. cone, 1986). DELSOL is based in part on the performances
Interesting features were added during solar two design design approaches developed at the University of Houston
and operation: an allowable flux constraint was used to (images generated by Hermite polynomials convolution,
optimize the new heliostat layout (Vant-Hull and Pitman, solar field divided into zones defined by their heliostat den-
1988) and a C++ version of UHC incorporating the struc- sity for optimization). DELSOL optimizes tower height,
ture and imaging characteristics of IH helped to compute, receiver dimensions, and field boundaries on an economic
in real-time, aiming strategies to control solar flux density basis, but requires a definition of the heliostat spacings in
on a receiver aperture (protection against excess flux den- the field.
sity) (Vant-Hull et al., 1996a). WINDELSOL is an adaptation of DELSOL for Win-
dows with new features (user-friendly interface, optimized
5.3.2. DELSOL/WINDELSOL defined-by-coordinates heliostat field generation. . .) but
This FORTRAN code was developed to fill the need for cannot be used to analyze existing facilities (heliostat coor-
an accurate, yet fast, easy to use tool for performance anal- dinates cannot be user-defined). Subdivision of field into
ysis, design and optimization of large and small power sys- sectors and data entry organization make it difficult to
tems for electricity and process heat applications (Kistler, introduce new optical parameters. This program seems to
1986). It typically requires less computer time for perfor- be adapted to large systems but, as its developers grant,
 (1986), and then applied to other concepts (like Phoebus).
It calculates the annual production of a CRS plant depend-
ing on its configuration (heliostats, tower, receiver, cycle).
The layout and optimization of the global system are
designed to maximize electricity production per heliostat
or minimize the cost of produced electricity. The annual
production is assessed with an error smaller than 5% in
comparison with the US codes HELIOS, DELSOL, and
UHC (Kiera, 1989). Nowadays it is still updated and used
for CRS modeling by DLR researchers (Final Technical
Report Solgate, 2005), who consider it easier to adapt than
DELSOL.
HFLCAL is able to carry out the hundreds of annual
performance calculations needed for determining a cost–
optimized solar field layout in a reasonable time. It uses
a simplified optical model with a flux distribution reflected
by each heliostat approached by a circular Gaussian distri-
bution obtained by convolution of three distributions (one
accounting for the size and the luminance of the sun, a sec-
ond for the heliostat error and a third one for tracking
errors). Conversely in the 4 other codes described here var-
ious mirror errors are detailed (curvature, canting, micro-
scopic. . .) panel by panel. Thus in HFLCAL it is
sufficient to consider the central ray of one heliostat and
to include these three errors with three standard deviations,
according to a model settled by Rabl (1985). Consequently
this code only does one calculation per heliostat whose
focal length is considered constant or equal to slant range.
HFLCAL was used during the SOLGATE project to
cost-optimize the solar part of a hybrid CRS plant with
pressurized air volumetric receiver. Heliostat field size
and layout, tower height, receiver geometry were adapted
to the specific needs of HSGT systems for a given capacity
at a given location. Comparisons with experimental mea-
Fig. 2. Visualization of results from Windelsol and Soltrace for the Pegase sures showed that total power on receiver aperture can eas-
facility: (a) annual performances of the solar field from WINDELSOL ily be calculated with high precision, whereas the precision
(view from above); (b) impacting rays on the collector surface from
SOLTRACE.
on maximum flux is low and the flux distribution is not
accurate (Final Technical Report Solgate, 2005).
not accurate for small plants. Its user-friendliness turns it
to an interesting tool for preliminary studies of large
CRS projects. 5.3.4. MIRVAL
This tool was developed for the rigorous optical perfor-
5.3.3. HFLCAL mance analysis of the envisaged concepts during Solar One
HFLCAL code (for ‘‘Heliostats Field Layout CALcula- preliminary study. It was used to check flux calculations
tions’’) was developed for the GAST hybrid concept from DELSOL, HELIOS and UHC. It is quite simple to
Table 2
Two categories of flux calculation codes
Optimization codes Performance analysis codes
Tools UHC-RCELL, (WIN)DELSOL, HFLCAL UHC-IH or NS, MIRVAL, FIAT LUX,
SOLTRACE, HELIOS, (DELSOL)
Considered subsystems Overall plant Optical subsystem
Data entry Total ground area of the solar field divided Detailed heliostat geometry, in text files.
into zones, in windows
Calculation method Simplified convolution Monte Carlo or similar
Calculation time Some seconds for a whole heliostat field Some seconds to some minutes for one single heliostat
Accuracy Increase with field size Accurate for one heliostat
Economic calculations Yes No
use and, as it is implemented with Monte Carlo method, 6. Other codes
ray vectors onto the receiver aperture can be provided, so
as to interface with a receiver or secondary concentrator The ray-tracing code OPTEC (for ‘‘optimization pro-
model. For Izygon et al. (1987), MIRVAL turned out to gram for terminal concentrators’’, Schoffel and Sizmann,
be unsuitable for modeling the complex geometry of 1991) can handle a solar furnace (Neumann and Witzke,
CETHEL heliostats of the Themis field. 1999), a heliostat field or a parabolic dish and add a cone,
trumpet or compound parabolic concentrator as terminal
5.3.5. FIAT LUX concentrator.
Initially designed to validate the optical quality of helio- Similarly to research centers, industrial groups promot-
stats, it enables the user to analyze in details existing instal- ing CSP projects took an interest in calculation of solar
lations, with a sharp representation of the heliostat field performances through the development of similar
geometry. Heliostat orientation can be easily checked tools, often dedicated to design optimization, and not dif-
before calculation by 3-D visualization of reflected rays fused. For example one can note SOLVER from SOLU-
from the center of the facets. Flux distribution after calcu- CAR (ray tracing for CRS modeling) and SENSOL from
lation is statistically described. FIAT LUX was compared SENER (for CRS and PT modeling, Relloso and Domin-
with HELIOS on a Martin-Marietta heliostat, showing go, 2006).
good agreement (Monterreal, 2000). Nevertheless this tool To complete this overview we shall also quote codes cur-
is now being developed: a few functionalities are still miss- rently in development, intended to replace codes from the
ing such as S&B effects and user interface. 80s, sometimes considered as hard to use and modify,
non-adapted to dynamic modeling, lacking documentation,
5.3.6. SOLTRACE and limited by obsolete computing power constraints
The objective of SOLTRACE is to model complex opti- (Blanco et al., 2005). These new codes, more modular, will
cal systems for solar power and to analyze their perfor- tackle more easily simulation of new CSP systems, such as
mance (Wendelin, 2003). It is adapted not only to the hybrid concepts, cogeneration plants or with multi-staged
three main technologies to thermodynamically convert receivers. Besides SOLTRACE which may constitute the
sun power into electricity (CRS, PT and DS) but also to first code of this generation, TONATIUH (Blanco et al.,
solar furnaces. The optical system is organized in stages 2005) aims to be a sophisticated software environment
in a global coordinates system: stages are sections of the for design, optimization and analysis of all CRS, PT, and
optical geometry that are successively hit by rays in their DS. This open source ray-tracing code will enable users
pathway from the sun to the final receiver. They can be to visualize the installation with state-of-the-art 3-D inter-
optical ones (physical interactions with rays) or virtual face technologies. CIEMAT laboratory is currently devel-
ones (useful to determine positions of rays or flux maps oping a series of tools called SCT (Solar Concentration
wherever in the system without interacting with them). Toolbox) Package, developed under MATLAB software
The reason for this organization is to trace rays in an (Sanchez and Romero, 2006). One tool is dedicated to opti-
efficient way in order to limit calculation time. One stage cal design and optimization of solar receivers, another one
is composed by elements with their own aperture, to generation of random rays, and the last one to CRS
shape, and optical properties (slope error, specularity optimization and performance. These codes will use either
error, reflectivity, transmissivity, and refraction). Each ele- ray-tracing or convolution techniques. At last the EPFL-
ment is described in a coordinates system related to its LENI laboratory worked to integrate a heliostat-tower
stage. model built with RADIANCE in a MATLAB environment
Contrary to other tools, atmospheric attenuation and and has recently developed a multi-objective evolutionary
tracking errors are not directly included but they can be algorithm for the design of heliostat fields (Pelet et al.,
taken into account, respectively, in the reflectivity of the 2006).
mirrors and in their slope error. Moreover the complete
description of geometry of the elements leads to a cumber-
some interface. Indeed for each system all the heliostat field 7. How to chose an adapted calculation code?
geometry must be built in a spreadsheet including:
For researchers or engineers that would need to model
– definition of the center, the aperture, the normal direc- CRS, it may be difficult to find the right tool that best fit
tion, and the curvature of each facet of each heliostat to the question asked. We hope that this study will help
in the stage coordinates system; them to solve this problem. In the case of an industrial pro-
– definition of each stage coordinates system in relation to ject, an interesting strategy may be to determine first the
the global coordinates system. general layout of the plant (tower height, heliostat field
boundary, receiver geometry and technology, storage
This organization implies the calculation of the position capacity) from key parameters such as power block charac-
of each facet of each heliostat depending on the sun teristics, meteorological data, and load curves. This analy-
position. sis can be done with tools from the first category. Then a
more detailed study including a closer description of helio- Acknowledgements
stats and receiver geometry can be led with tools from the
second category to assess solar field performances or to The authors wish to thank Allan Lewandowski from
improve pointing and operating strategies. NREL, Marcelino Sanchez and Rafael Monterreal from
Another approach to assess global performances of CIEMAT, Mark Schmitz, Peter Schwarzbözl, and Reiner
CRS may be the following: Buck from DLR, François Hénault from CRAL - Observa-
toire de Lyon, Manuel Blanco from University of Texas,
– first to use solar field efficiency matrixes and receiver Lorin Vant-Hull from University of Houston and Rafaele
performances curves from specific codes like the five Bolliger from EPFL-LENI for their invaluable help.
ones quoted above,
– then to integrate these data in TRNSYS with STEC
library or in less detailed tools based on a simple energy References
balance formulation (SOLERGY, ECOSTAR).
Biggs, F., Vittitoe, C.N., 1977. HELIOS: a computational model for solar
concentrators, Sandia National Labs, Albuquerque, NM, US-USSR
With this second approach, one component can be mod- Workshop on Solar Energy Appl., Moscow.
ified more easily without changing the global model. Blanco, M.J., Amieva, J.M., Mancilla, A., 2005. The TONATIUH
Besides a wider range of concepts can be simulated. How- software development project: an open source approach to the
ever efforts must be made to link components in a consis- simulation of solar concentrating systems. In: Proceedings of 2005
ASME International Mechanical Engineering Congress and Exposi-
tent way to avoid discrepancies.
tion, Orlando, Florida.
Falcone, P.K., 1986. A Handbook for Solar Central Receiver Design,
Sandia Report SAND86-8009.
8. Conclusions and perspectives Final Publishable Report Solgate, solar hybrid gas turbine electric power
system, 2005. Contract ENK5-CT-2000-00333, Project Partly Funded
by the European Community in the Framework of the 5th RTD
This study has permitted us to classify tools into two
Framework Programme 1998–2002. Partners: Ormat, Ciemat, DLR,
categories, corresponding to two types of problems Solucar, Tuma, http://ec.europa.eu/research/energy/pdf/solgate_en.
encountered in studies on CSP technologies: pdf.
Garcia, P., Ferriere, A., Flamant, G., Costerg, P., Soler, R., Gagnepain,
(1) optimization codes to design installation projects like B., 2006. The PEGASE project: design and performance evaluation of
a hybrid solar/fossil gas turbine system for electricity generation. In:
HFLCAL, UHC-RCELL, or (WIN)DELSOL;
Romero, M., Martı́nez, D., Ruı́z, V., Silva, M., Brown, M. (Eds.),
(2) performance analysis codes to assess optical perfor- Proceedings of SolarPACES 13th International Symposium on Con-
mances of a given installation like FIAT LUX, MIR- centrated Solar Power and Chemical Energy Technologies, June 20–
VAL, UHC-NS or IH, or SOLTRACE. 23, Seville, Spain, pp. A5–S6.
Hénault, F., 1987. Concentration du rayonnement solaire par simple et
double réflexion: contribution aux méthodes de réglage et de contrôle
It is worth noting that other types of modeling tools
des surfaces réflectrices à facettes, thèse soutenue à l’Université de
might be necessary to design and study CRS systems like Paris-Sud, pp. 35–60.
specific codes for receiver modeling, for heliostat field lay- Hénault, F., 2005. COSAC Version 2.5, Présentation Générale, http://
out optimization or transient analysis. francois.henault.free.fr/cosac/cosac252_1.pdf.
For research laboratories or industrial developers, it Izygon, M., Bezian, J.J., Hénault, F., 1987. Modélisation des flux réfléchis
par un champ d’héliostats, Comparaison de divers modèles. Proceed-
seems interesting to use one code from each category.
ings of Séminaire héliothermie. Springer, NY.
Indeed codes from the first category are able to assess the Kiera, M., 1986. Beschreibung und Handhabung des Programmsystems
available flux distribution at the top of a tower and thus HFLCAL, Interatom Report.
to size up a receiver. That is why we chose to use SOL- Kiera, M., 1989. Heliostat field: computer codes, requirements, compar-
TRACE to model the optical components of an existing ison of methods. In: Becker, M., Böhmer, M. (Eds.), Proceedings of
the Final GAST Presentation. Springer, Lahnstein, Germany, pp. 95–
facility. This code will be used to study the solar field per-
113.
formances depending on pointing strategy, to describe the Kistler, B.L., 1986. A user’s manual for DELSOL3: a computer code for
flux distribution on receiver module aperture, and to elab- calculating the optical performance and optimal system design for
orate an accurate efficiency matrix. Besides SOLTRACE solar thermal central receiver plants, Sandia Report SAND86-8018.
can be adapted to other concentrating systems (solar fur- Laurence, C.L., Lipps, F.W., Vant-Hull, L.L., 1984. User’s manual for the
University of Houston individual heliostat layout and performance
naces, dish-Stirling). Tools from the second category, like
code. Houston University, TX.
(WIN)DELSOL, are recommended for techno-economic Leary, P.L.; Hankins, J.D., 1979. Users guide for MIRVAL: a computer
assessment of CRS technologies. However various addi- code for comparing designs of heliostat-receiver optics for central
tions or modifications must be brought to adapt them to receiver solar power plants, Sandia Report SAND-77-8280.
new CRS concepts. Lipps, F.W., Vant-Hull, L.L., 1978. A Cellwise method for the optimi-
zation of large central receiver systems. Sol. Energ 20, 505–516.
Comparing results of updated codes from both catego-
Lipps, F.W., Vant-Hull, L.L., 1980a. Programmer’s manual for the
ries on one or various reference test cases should be a chal- University of Houston computer code RCELL: cellwise optimization
lenging work in partnership for research centers involved in for the solar central receiver project, Technical Report Number(s)
CRS or other CSP technologies. DOE/SF/10763-T5, SAND-0763-1.
Lipps, F.W., Vant-Hull, L.L., 1980b. In: User manual for the University centrated Solar Power and Chemical Energy Technologies, June 20–
of Houston solar central receiver, cellwise performance model: NS, vol. 23, Seville, Spain, pp. A7–S8.
1. Houston University, TX. Sanchez, M., Romero, M., 2006. Methodology for generation of heliostat
Matteı̈, S. 2005. Rayonnement thermique des matériaux opaques, Tech- field layout in central receiver systems based on yearly normalized
niques de l’ingénieur, file BE8210. energy surfaces. Solar Energy 80, 861–874.
Monterreal, R., 2000. Software developments for system analysis and Schoffel, U., Sizmann, R., 1991. Final report terminal concentrators. Solar
optimization. SolarPACES Tech. Report No. III-3/00. In: Pitz-Paal, Thermal Energy Utilization. Springer, Berlin, p. 183, 3-53268-4.
R., Schwarzbözl, P. (Eds.), Proc. of the IEA SolarPACES Task III Schwarzbözl, P., Eiden, U., Pitz-Paal, R., 2002. A TRNSYS model library
Workshop on Simulation of Solar Thermal Power Systems 2000, for solar thermal electric components (STEC), a reference manual
Cologne, Germany. release 2.2. IEA-solar power and chemical energy systems, Task III:
Neumann, A., Witzke, A., 1999. The influence of sunshape on the DLR solar technologies and applications, http://sel.me.wisc.edu/trnsys/trn-
solar furnace beam. Sol. Energy 66, 447–457. lib/stec/stec_2.2_refguide.pdf.
Pelet, X., Favrat, D., Sanchez, M., Romero, M., 2006. Design of heliostat Stoddard, M.C., Faas, S.E., Chiang, C.J., Dirks, A.J., 1987. SOLERGY –
fields using a multi-objective evolutionary algorithm. In: Romero, M., A Computer Code for Calculating the Annual Energy from Central
Martı́nez, D., Ruı́z, V., Silva, M., Brown, M. (Eds.), Proceedings of Receiver Power Plants. SAND86-8060. Sandia National Laboratories,
SolarPACES 13th International Symposium on Concentrated Solar Livermore, CA.
Power and Chemical Energy Technologies, June 20–23, Seville, Spain, Vant-Hull, L.L., Pitman, C.L., 1988. Central receiver system optimization
pp. A7–S1. under an allowable flux constraint. In: Gupta, B.P., Traugott, W.H.
Pitman, C.L., Vant-Hull, L.L., 1989. The University of Houston solar (Eds.), Proceedings 4th International Symposium on Solar Thermal
central receiver code system: concepts, updates and start-up kits, Technology-Research, Development and Applications, Santa Fe, NM.
SAND88-7029, Houston TX, University of Houston. Hemisphere Publishing Corporation, New York, pp. 51–60.
Pitz-Paal, R., Schwarzbözl, P. 2000. Software developments for system Vant-Hull, L.L., Izygon, M.E., Pitman, C.L., 1996a. Real time compu-
analysis and optimization. SolarPACES Tech. Report No. III-3/00. In: tation and control of solar flux density on a central receiver (solar two )
Proceedings of the IEA SolarPACES Task III Workshop on Simula- (protection against excess flux density). In: Campbell-Howe, R.,
tion of Solar Thermal Power Systems, Cologne, Germany. Wilkins-Crowder, B. (Eds.), Proceedings of the ASES Annual Con-
Pitz-Paal R., Dersch, J., Milow, B., Ferriere, A., Romero, M., Tellez, F., ference. Asheville, NC.
Steinfeld, A., Langnickel, U., Shpilrain, E., Popel, O., Epstein, M., Walzel, M.D., Lipps, F.W., Vant-Hull, L.L., 1977. A solar flux density
Karni, J., 2005. ECOSTAR Road Map Document, SES-CT-2003- calculation for a solar tower concentrator using a two-dimensional
502578, ftp://ftp.dlr.de/ecostar/. hermite function expansion. Solar Energy 19 (3), 239–253.
Rabl, A., 1985. Active Solar Concentrators and Their Applications. Wendelin, T., 2003. SolTRACE: a new optical modeling tool for
Oxford University Press, New York. concentrating solar optics. In: Proceedings of the ISEC 2003: Inter-
Relloso, S., Domingo, M., 2006. Solar projects analysis using SENSOL. national Solar Energy Conference, 15–18 March 2003, Kohala Coast,
In: Romero, M., Martı́nez, D., Ruı́z, V., Silva, M., Brown, M. (Eds.), Hawaii, New York, American Society of Mechanical Engineers, pp.
Proceedings of SolarPACES 13th International Symposium on Con- 253–260.