Department ofb School of Engic Facultad de In

24 January 2013

Keywords:

Concentrated solar power plants

Hourly beam irradiation

Plant simulation

ar radiation and energy demand peak, CSPs experience short term variations

t provide energy during night hours unless incorporating thermal energy

storage (TES) and/or backup systems (BS) to operate continuously. To determine the optimum design

and operation of the CSP throughout the year, whilst dening the required TES and/or BS, an accurate

estimation of the daily solar irradiation is needed. Local solar irradiation data are mostly only available

. . . . . .

ide en

plants

. . . . . .

. . . . . .

. 472

. 472

. 473

5. Computing global and diffuse solar hourly irradiation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474

Contents lists available at SciVerse ScienceDirect

Renewable and Sustainable Energy Reviews

Renewable and Sustainable Energy Reviews 22 (2013) 466481E-mail address: Zhanghl.lily@gmail.com (H.L. Zhang).5.2.1. Estimating the daily irradiation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474

5.2.2. Sequence of days . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475

1364-0321/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.rser.2013.01.032

n Corresponding author. Tel.: 32 16 322695; fax: 32 16 322991.5.1. Background information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474

5.2. The adopted model approach and equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4744. Enhancing the CSP potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.1. Thermal energy storage systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.2. Backup systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.1.1. Solar power towers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468

2.1.2. Parabolic trough collector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469

2.1.3. Linear Fresnel reector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470

2.1.4. Parabolic dish systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470

2.1.5. Concentrated solar thermo-electrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471

2.2. Comparison of CSP technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471

3. Past and current SPT developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472Contents

1. Introduction . . . . . . . . . . . . . . . .

1.1. Solar irradiance as worldw

1.2. Concentrated solar power

2. CSP technologies . . . . . . . . . . . .

2.1. Generalities . . . . . . . . . .as monthly averages, and a predictive conversion into hourly data and direct irradiation is needed to

provide a more accurate input into the CSP design. The paper (i) briey reviews CSP technologies and

STC advantages; (ii) presents a methodology to predict hourly beam (direct) irradiation from available

monthly averages, based upon combined previous literature ndings and available meteorological data;

(iii) illustrates predictions for different selected STC locations; and nally (iv) describes the use of the

predictions in simulating the required plant conguration of an optimum STC.

The methodology and results demonstrate the potential of CSPs in general, whilst also dening the

design background of STC plants.

& 2013 Elsevier Ltd. All rights reserved.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467

ergy source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467Design methodology

Solar towersAccepted 26 January 2013hours of the day where sol

on cloudy days and cannoa r t i c l e i n f o

Article history:

Received 17 November 2012

Received in revised form

a b s t r a c t

Concentrated solar power plants (CSPs) are gaining increasing interest, mostly as parabolic trough

collectors (PTC) or solar tower collectors (STC). Notwithstanding CSP benets, the daily and monthly

variation of the solar irradiation ux is a main drawback. Despite the approximate match betweenChemical Engineering, Chemical and Biochemical Process Technology and Control Section, Katholieke Universiteit Leuven, Heverlee 3001, Belgium

neering, University of Warwick, Coventry, UK

geniera y Ciencias, Universidad Adolfo Ibanez, Santiago, ChileConcentrated solar power plants: Review and design methodology

H.L. Zhang a,n, J. Baeyens b, J. Degreve a, G. Caceres c

ajournal homepage: www.elsevier.com/locate/rser

n .

es .

. . .

. . .

. . .

. . .

. . .

. . .

. . .

. . .

. . .

dening thetion of the dfor worldwi

vious theoretical and experimental ndings into a general method

calculations from the temperatures recorded at the locations;

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481 467capacity of TES and required BS, an accurate estima-aily solar irradiation is needed. Solar irradiation data

tionally, a recent technology called concentrated solar thermo-electrics is described. These CSP technologies are currently inexperience short term variations on cloudy days and cannotprovide energy during night hours. In order to improve the overallyield in comparison with conventional systems, the CSP processcan be enhanced by the incorporation of two technologies, i.e.,thermal energy storage (TES) and backup systems (BS). Bothsystems facilitate a successful continuous and year round opera-tion, thus providing a stable energy supply in response toelectricity grid demands. To determine the optimum design andoperation of the CSP throughout the year, whilst additionally

water/steam) or via a secondary circuit to power a turbine andgenerate electricity. CSP is particularly promising in regions withhigh DNI. According to the available technology roadmap [8], CSPcan be a competitive source of bulk power in peak and inter-mediate loads in the sunniest regions by 2020, and of base loadpower by 2025 to 2030.

At present, there are four available CSP technologies (Fig. 2):parabolic trough collector (PTC), solar power tower (SPT), linearFresnel reector (LFR) and parabolic dish systems (PDS). Addi-problem for all CSP technologies: despite the close matchbetween hours of the day in which energy demand peaks andsolar irradiation is available, conventional CSP technologies

trated on a small area. Using mirrors, sunlight is reected to areceiver where heat is collected by a thermal energy carrier(primary circuit), and subsequently used directly (in the case of5.2.3. Estimation of the hourly diffuse and beam radiatio

5.2.4. Shortcut estimates, based on recorded temperatur

6. Model parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6.1. Common measurement methods of solar radiation . . . . . . .

6.2. Available information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6.3. Selected locations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7.1. Calculations of H0, H and Hb . . . . . . . . . . . . . . . . . . . . . . . . .

7.2. Methodology to apply the predictions in CSP design . . . . . .

8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction

1.1. Solar irradiance as worldwide energy source

More energy from the sunlight strikes the earth in 1 h than allof the energy consumed by humans in an entire year. In fact, solarenergy dwarfs all other renewable and fossil-based energyresources combined.

We need energy electrical or thermal but in most caseswhere and when it is not available. Low cost, fossil-basedelectricity has always served as a signicant cost competitor forelectrical power generation. To provide a durable and widespreadprimary energy source, solar energy must be captured, stored andused in a cost-effective fashion.

Solar energy is of unsteady nature, both within the day (daynight, clouds) and within the year (wintersummer). The captureand storage of solar energy is critical if a signicant portion of thetotal energy demand needs to be provided by solar energy.

Fig. 1 illustrates the world solar energy map. Most of thecountries, except those above latitude 451N or below latitude451S, are subject to an annual average irradiation ux in excess of1.6 MW h/m2, with peaks of solar energy recorded in some hotspots of the Globe, e.g., the Mojave Desert (USA), the Sahara andKalahari Deserts (Africa), the Middle East, the Chilean AtacamaDesert and North-western Australia.

1.2. Concentrated solar power plants

Concentrated solar power plants are gaining increasing interest,mostly by using the parabolic trough collector system (PTC),although solar power towers (SPT) progressively occupy a signi-cant market position due to their advantages in terms of higherefciency, lower operating costs and good scale-up potential. Thelarge-scale STC technology was successfully demonstrated byTorresol in the Spanish Gemasolar project on a 19.9 MWel-scale [2].

Notwithstanding CSP benets, the varying solar radiation uxthroughout the day and throughout the year remains a mainde locations are mostly only available as monthly select an appropriate plant conguration, and present designpreliminary recommendations using predicted hourly beamirradiation data.

In general, the study will demonstrate the global potential ofimplementing the SPT technology, and will help to determine themost suitable locations for the installation of SPT plants.

2. CSP technologies

2.1. Generalities

Concentrated solar power (CSP) is an electricity generationtechnology that uses heat provided by solar irradiation concen-meestimate the hourly beam irradiation ux from availablemonthly mean global irradiation data for selected locations,and compare the results obtained of monthly data withreview the CSP technologies and discuss solar power toweradvantages compared to the other technologies;of calculating the hourly beam irradiation ux. The basis waspreviously outlined by Dufe and Beckmann [3], and uses the Liuand Jordan [4] generalized distributions of cloudy and clear days,later modied by Bendt et al. [5], then by Stuart and Hollands [6]and nally by Knight et al. [7].

The present paper has therefore the following specicobjectives:. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480

averages, and a predictive conversion into hourly data and directirradiation is needed to provide a more accurate input into theCSP design. Considering that a CSP plant will only accept directnormal irradiance (DNI) in order to operate, a clear day model isrequired for calculating the suitable irradiation data.

The procedure, outlined in the present paper, combines pre-dium to large-scale operation and mostly located in Spain and

Nomenclature

Abbreviations

BS Backup systemCRS Central receiver systemCSP Concentrated solar power plantCLFR Compact linear Fresnel collectorDNI Direct normal irradianceDSG Direct steam generationHCE Heat collector elementHFC Heliostat eld collectorHTF Heat transfer uidISCC Integrated solar combined cycleLFR Linear Fresnel reectorNREL National Renewable Energy LaboratoryPDC Parabolic dish collectorPTC Parabolic trough collectorTES Thermal energy storageS&L Sargent and LundySNL Sandia National LaboratoriesSTC Solar tower collector

radiation, at mean earth-sun distance, outside of theatmosphere

H0 the extra-terrestrial radiation (MJ/m2 day)

Ho,av The monthly average of H0H The daily total radiation obtained from the registered

measurementsHav The monthly average of HHd The daily diffuse radiationI The hourly radiationId The hourly solar diffuse radiationIb The hourly solar beam radiationI0 The hourly extraterrestrial radiationKT,av Monthly average clearness indexKT Daily clearness index;kT Hourly clearness index;KT,min Minimum daily clearness indexKT,max Maximum daily clearness indexKRS Hargreaves adjustment coefcient (1C

0.5) (0.16/0.19)n The nth-day of the yearn Number of the day of the month (1, 2,y nd )

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481468Symbols

a Parameter dened by Eq. (17)in the USA as shown in Fig. 3. Although PTC technology is themost mature CSP design, solar tower technology occupies thesecond place and is of increasing importance as a result of itsadvantages, as discussed further.

Fig. 1. World solar e

b Parameter dened by Eq. (18)dr The inverse relative distance EarthSunF Cumulative distribution function or fraction of days in

which the daily clearness index in less than a certainspecic value;

GSC the solar constant1367 W/m2, as energy of the sunper unit time received on a unit area of the surfaceperpendicular to the propagation direction of thedk k

ndm Number of the days in a certain month (31, 30 or 28)rt The ratio of hourly to total radiationrd The ratio of hourly diffuse to daily diffuse radiationTmax Maximum air temperature (1C)Tmin Minimum air temperature (1C)ws The sunset hour angle (rad)2.1.1. Solar power towers

Solar power towers (SPT), also known as central receiversystems (CRS), use a heliostat eld collector (HFC), i.e., a eld ofsun tracking reectors, called heliostats, that reect and concen-trate the sunrays onto a central receiver placed in the top of axed tower [2,9]. Heliostats are at or slightly concave mirrors

nergy map [1].

w The hour angle of the sun (rad)d The solar declination angle (rad)g Parameter that denes the exponential distribution

proved by Bouguer law of absorption of radiationthrough the atmosphere

Latitude of the location (rad)x Dimensionless parameter, dened by Eq. (9)

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481 469that follow the sun in a two axis tracking. In the central receiver,heat is absorbed by a heat transfer uid (HTF), which thentransfers heat to heat exchangers that power a steam Rankinepower cycle. Some commercial tower plants now in operation usedirect steam generation (DSG), others use different uids, includ-ing molten salts as HTF and storage medium [9]. The concentrat-ing power of the tower concept achieves very high temperatures,thereby increasing the efciency at which heat is converted intoelectricity and reducing the cost of thermal storage. In addition,the concept is highly exible, where designers can choose from awide variety of heliostats, receivers and transfer uids. Someplants can have several towers to feed one power block.

2.1.2. Parabolic trough collector

A parabolic trough collector (PTC) plant consists of a group ofreectors (usually silvered acrylic) that are curved in one dimen-sion in a parabolic shape to focus sunrays onto an absorber tubethat is mounted in the focal line of the parabola. The reectorsand the absorber tubes move in tandem with the sun as it dailycrosses the sky, from sunrise to sunset [9,10]. The group ofparallel connected reectors is called the solar eld.

Typically, thermal uids are used as primary HTF, thereafterpowering a secondary steam circuit and Rankine power cycle.

Fig. 2. Currently available CSP Technologie

USA40.1%

Spain57.9%

Iran1.4% Italy

0.4%

Australia 0.2%

Germany0.1%

Fig. 3. Installed operational CSP power (MarchOther congurations use molten salts as HTF and others use adirect steam generation (DSG) system.

The absorber tube (Fig. 4), also called heat collector element(HCE), is a metal tube and a glass envelope covering it, with eitherair or vacuum between these two to reduce convective heat lossesand allow for thermal expansion. The metal tube is coated with a

s:(a) STP; (b)PTC; (c) LFR; (d) PDC [8].

Parabolic Trough96.3%

Solar Tower3.0%

Parabolic Dish0.1%

Linear Fresnel0.7%

2011), by country and by technology [10].

Fig. 4. Absorber element of a parabolic trough collector [9].

selective material that has high solar irradiation absorbance andlow thermal remittance. The glass-metal seal is crucial in redu-cing heat losses.

2.1.3. Linear Fresnel reector

Linear Fresnel reectors (LFR) approximate the parabolic shapeof the trough systems by using long rows of at or slightly curvedmirrors to reect the sunrays onto a downward facing linearreceiver. The receiver is a xed structure mounted over a towerabove and along the linear reectors. The reectors are mirrorsthat can follow the sun on a single or dual axis regime. The mainadvantage of LFR systems is that their simple design of exiblybent mirrors and xed receivers requires lower investment costsand facilitates direct steam generation, thereby eliminating theneed of heat transfer uids and heat exchangers. LFR plants arehowever less efcient than PTC and SPT in converting solar energyto electricity. It is moreover more difcult to incorporate storagecapacity into their design.

A more recent design, known as compact linear Fresnelreectors (CLFR), uses two parallel receivers for each row of

mirrors and thus needs less land than parabolic troughs toproduce a given output [11].The rst of the currently operatingLFR plants, Puerto Errado 1 plant (PE 1), was constructed inGermany in March 2009, with a capacity of 1.4 MW. The successof this plant motivated the design of PE 2, a 30 MW plant to beconstructed in Spain. A 5 MW plant has recently been constructedin California, USA.

2.1.4. Parabolic dish systems

Parabolic dish collectors (PDC), concentrate the sunrays at afocal point supported above the center of the dish. The entiresystem tracks the sun, with the dish and receiver moving intandem. This design eliminates the need for a HTF and for coolingwater. PDCs offer the highest transformation efciency of any CSPsystem. PDCs are expensive and have a low compatibility withrespect of thermal storage and hybridization [11]. Promotersclaim that mass production will allow dishes to compete withlarger solar thermal systems [11]. Each parabolic dish has a lowpower capacity (typically tens of kW or smaller), and each dishproduces electricity independently, which means that hundreds

Fig. 5. Concentrated solar thermo-electric technology[11].

yna

Mean gross efciency (as % of direct radiation) 15.4

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481470Mean net efciency (%) 14

Specic power generation (kW h/m2-year) 308

Capacity factor (%) 2350

Unitary investment (h/kW hel) 1.54

Levelized electricity cost (h/kW hel) 0.160.19Table 1Comparison between leading CSP technologies [8,11,13].

Relative cost Land occupancy Cooling water

(L/MW h)

Thermo-d

efciency

PTC Low Large 3,000 or dry Low

LFR Very low Medium 3,000 or dry Low

SPT High Medium 1,500 or dry High

PDC Very high Small None High

Table 2Comparison for 50 MWel CSP plants with TES.

ParametersPTC with oil, without

storage and back-upSPT with steam, without

storage and back-up

SPT with molten salt, TES storage

and back-up system

14.2 18.1

13.6 14

258 375

24 Up to 75

1.43 1.29

0.170.23 0.140.17mic Operating

T range (1C)Solar concentration

ratio

Outlook for improvements

20400 1545 Limited

50300 1040 Signicant

300565 1501500 Very signicant

1201500 1001000 High potential through

mass production

or thousands of them are required to install a large scale plant likebuilt with other CSP technologies [11].

Maricopa Solar Project is the only operational PDC plant, witha net capacity of 1.5 MW. The plant began operation on January2010 and is located in Arizona, USA.

2.1.5. Concentrated solar thermo-electrics

As well as with photovoltaic systems, direct conversion ofsolar energy into electricity can also be achieved with concen-trated solar thermo-electric (CST) technology. Solar thermo-electric devices can convert solar thermal energy, with its inducedtemperature gradient, into electricity. They can also be modiedto be used as a cooling or heating technology [11]. Recently, CSPtechnologies have been combined with thermo-electrics in orderto achieve higher efciencies [11]. A concentrated solar thermo-electric power generator typically consists of a solar thermalcollector and a thermo-electric generator (Fig. 5). Heat isabsorbed by the thermal collector, then concentrated and con-ducted to the thermo-electric generator, where the thermalresistance of the generator creates a temperature differencebetween the absorber plate and the uid, which is proportionalto the heat ux. The current cost of thermo-electric materialshampers the widespread use of CSTs.

2.2. Comparison of CSP technologies

Within the commercial CSP technologies, parabolic troughcollector (PTC) plants are the most developed of all commerciallyoperating plants [12]. Table 1 compares the technologies on thebasis of different parameters.

In terms of cost related to plant development, SPT and PDCsystems are currently more expensive, although future

developments and improvements [13] will alter levelized energycost projections, as presented by Sandia National Laboratories(SNL) and by Sargent & Lundy Consulting Group (S&L): SPT will bethe cheaper CSP technology in 2020.

In terms of land occupancy, considering the latest improve-ments in CSP technologies, SPT and LFR require less land than PTCto produce a given output. Additionally, PDC has the smallest landrequirement among CSP technologies [8,12].

Water requirements are of high importance for those locationswith water scarcity, e.g., in most of the deserts. As in otherthermal power generation plants, CSP requires water for coolingand condensing processes, where requirements are relativelyhigh: about 3000 L/MW h for PTC and LFR plants (similar to anuclear reactor) compared to about 2000 L/MW h for a coal-redpower plant and only 800 L/MW h for a combined-cycle naturalgas power plant. SPT plants need less water than PTC (1500 L/MW h) [8]. Dishes are cooled by the surrounding air, so they donot require cooling water. Dry cooling (with air) is an effective

Hea

Liqu

Stea

Stea

Stea

Stea

Mol

Hi-T

Stea

Air

Mol

Pres

Pres

Sup

Air

Wat

Air

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481 471Table 3PS20, Sierra sun tower and Gemasolar technical parameters [19].

Characteristics PS 20 Sierra sun tower Gemasolar

Turbine net capacity 20 MWel 5 MWel 19.9 MWelSolar eld area 150,000 m2 27,670 m2 304,750 m2

Number of heliostats 1,255 24,360 2,650

Heat transfer uid Water Water Molten salt

Receiver outlet temperature 2,550300 1C 440 1C 565 1CBackup fuel Natural gas Natural gas Natural gas

Storage capacity 1 h (No storage)15 h

(molten salt)

Capacity factor Approx. 27% Approx. 30% 7075%

Table 4Experimental solar power towers [12].

Project Country Power

PSA SSPS-CRS Spain 0.5 MWelEURELIOS Italy 1 MWelSUNSHINE Japan 1 MWelSolar One USA 10 MWelPSA CESA-1 Spain 1 MWelMSEE/Cat B USA 1 MWelTHEMIS France 2.5 MWelSPP-5 Russia 5 MWelTSA Spain 1 MWelSolar Two USA 10 MWelConsolar Israel 0.5 MWthSolgate Spain 0.3 MWelEureka Spain 2 MWelJulich Germany 1.5 MWelCSIRO SolarGas Australia 0.5 MWthCSIRO Brayton Australia 1 MWthalternative as proven by the plants under construction in NorthAfrica [8]. However, it is more costly and reduces efciencies. Drycooling systems installed on PTC plants located in hot deserts,reduce annual electricity production by 7% and increase the costof the produced electricity by about 10% [8]. However, theefciency reduction caused by dry cooling is lower for SPT thanfor PTC. The installation of hybrid wet and dry cooling systemsreduces water consumption while minimizing the performancepenalty. As water cooling is more effective, operators of hybridsystems tend to use only dry cooling in the winter when coolingneeds are lower, then switch to combined wet and dry coolingduring the summer.

A higher concentrating ratio of the sun enables the possibilityto reach higher working temperatures and better thermodynamicefciencies. On SPT plants, the large amount of irradiation focusedon a single receiver (2001000 kW/m2) minimizes heat losses,simplies heat transport and reduces costs [13].

In terms of technology outlooks, SPT shows promisingadvances, with novel HTF being developed and achieving highertemperatures to improve the power cycle efciencies. Moreover,higher efciencies reduce the cooling water consumption, andhigher temperatures can considerably reduce storage costs.

A tentative comparison of 50 MWel CSP plants with TES [13,14]is presented in Table 2. The capacity factor is dened as the ratioof the actual output over a year and its potential output if theplant had been operated at full nameplate capacity. Capacityfactors of CSP-plants without storage and back-up systems arealways low, due to the lacking power production after sunset andbefore sunrise.

t transfer uid Storage medium Operating since

id sodium Sodium 1981

m Nitrate salt/water 1981

m Nitrate salt/water 1981

m Oil/rock 1982

m Nitrate salt 1983

ten nitrate Nitrate salt 1984

ec salt Hi-Tec salt 1984

m Water/steam 1986

Ceramic 1993

ten nitrate Nitrate salt 1996

surized air Fossil hybrid 2001

surized air Fossil hybrid 2002

erheated steam Pressurized H2O 2009

Air/ceramic 2009

er/gas 2005

2011

A lower cost in SPT technology is mainly due to a lowerthermal energy storage costs, which benets from a largertemperature rise in the SPT compared to the PTC systems[14,15]. A higher annual capacity factor and efciency in SPT ismainly possible due to the thermal storage, which enables acontinuous and steady day-night output [14,16].

Additionally, in SPT plants, the whole piping system is con-centrated in the central area of the plant, which reduces the sizeof the piping system, and consequently reduces energy losses,material costs and maintenance [2,8]. In this scenario, solartowers with molten salt technology could be the best alternativeto parabolic trough solar power plants. Considering all mentionedaspects, SPT has several potential advantages. For both SPT andPTC technology, abundant quality data of main specic compo-nents are known [3,12,17,18], thus facilitating a more accurateanalysis of the technology.

3. Past and current SPT developments

The early developments included the PS 10 and a slightlyimproved PS 20 (Planta Solar 10 and 20) [18] of respectivecapacities 11 and 20 MWel, built near Sevilla. The plant technol-ogies involve glass-metal heliostats, a water thermal energystorage system (1 h), and cooling towers. A natural gas back-upis present [18,19]. The Sierra Sun Tower is the third commercialSPT plant in the world, and the rst of the United States. It consistof two modules with towers of 55 m height, total net turbinecapacity of 5 MWel and constructed on approximately 8 ha. Itbegan production in July 2009. Gemasolar is the fourth andnewest commercial SPT plant in the world, as it began productionin April 2011. It is the rst commercially operating plant to apply

located on 185 ha near Sevilla, Spain. The molten salt energystorage system is capable of providing 15 h of electricity produc-tion without sunlight, which enables the plant to provide elec-tricity for 24 consecutive hours. Table 3 shows the maincharacteristics of the PS 20, Sierra Sun Tower and Gemasolar SPT.

Additional pilot-SPT plants have been built and developedaround the world since 1981, as illustrated in Table 4 [12].

Commercial SPT plants are also being implemented, either inthe design or in the construction phase, as illustrated in Table 5.Recently additional large-scale projects have been announced fore.g., Morocco, Chile, the USA, and the Republic of South Africa.(RSA). The RSA announced an initiative of 5000 MW [20]. Theseprojects are not considered in Table 5, for current lack of detailedinformation.

4. Enhancing the CSP potential

As stated before, the CSP potential can be enhanced by theincorporation of two technologies in order to improve thecompetitiveness towards conventional systems: Thermal energystorage (TES) and backup systems (BS). Both systems offer thepossibility of a successful year round operation, providing a stableenergy supply in response to electricity grid demands [2,3].

4.1. Thermal energy storage systems

Thermal energy storage systems (TES) apply a simple princi-ple: excess heat collected in the solar eld is sent to a heatexchanger and warms the heat transfer uid (HTF) going from thecold tank to the hot tank. When needed, the heat from the hottank can be returned to the HTF and sent to the steam generator

l po

el

el

el

el

el

el

el

el

el

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481472molten salts as heat transfer uid and storage medium [2,19]. It is

Table 5Developing solar power tower projects [19].

Project Country Nomina

BrightSource Coyote springs 1 Nevada, USA 200 MW

BrightSource Coyote springs 2 Nevada, USA 200 MW

BrightSource PG&E 5 California, USA 200 MW

BrightSource PG&E 6 California, USA 200 MW

BrightSource PG&E 7 California, USA 200 MW

Crescent Dunes Solar Energy Project (Tonopah) Nevada, USA 110 MW

Gaskell sun tower California, USA 245 MW

Ivanpah Solar Electric Generating Station (ISEGS) California, USA 370 MW

Rice Solar Energy Project (RSEP) California, USA 150 MWFig. 6. Thermal energy storage system in(Fig. 6). In the absence of storage capacity, on the sunniest hours,

wer output HTF Storage medium Projected to start operation

Water July 2014

Water July 2015

Water July 2016

Water December 2016

Water July 2017

Molten salt Molten salt October 2013

Water May 2012

Water October 2013

Molten salt Molten salt October 2013a parabolic trough collector plant [8].

plant operators defocus some unneeded solar collectors to avoidoverheating the HTF. Storage avoids losing the daytime surplusenergy while extending the production after sunset.

Two types of thermal storage are necessary to maintain aconstant supply through the year, Short and Long term energystorage. Short term thermal energy storage collects and storessurplus daytime energy for nighttime consumption. Long termthermal energy storage is less obvious, since involving storage inspring and summer for autumn and winter months. Currently,only sensible heat is stored. The signicant improvement by usinglatent heat storage (phase change materials) or even chemicalheat storage (reversible endothermic/exothermic synthesis) is infull development [21], with chemical heat being considered moresuitable for long term thermal energy storage.

and to guarantee a nearly constant generation capacity, especiallyin peak periods. CSP plants equipped with backup systems arecalled hybrid plants. Burners can provide energy to the HTF, to thestorage medium, or directly to the power block. The integration ofthe BS can moreover reduce investments in reserve solar eld andstorage capacity. CSP can also be used in a hybrid mode by addinga small solar eld to a fossil fuel red power plant. These systemsare called integrated solar combined cycle plants (ISCC). As thesolar share is limited, such hybridization only limits fuel use. Apositive aspect of solar fuel savers is their relatively low cost:

Fig. 7. Heat capacity of different storage materials, (kW h/m3) versus meltingpoints (1K) [11].

Fig. 8. Cost of different storage materials (US$/kW h) versus melting points (1K) [10].

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481 473Thermal storage can be achieved directly or indirectly. Liquidse.g., mineral oil, synthetic oil, silicone oil, molten salts, can beused for sensible heat in direct thermal storage systems. Formolten salts, the desired characteristics for sensible heat usageare high density, low vapor pressure, moderate specic heat, lowchemical reactivity and low cost [21]. Indirect storage is whereHTF circulates heat, collected in the absorbers, and then pumpedto the thermal energy storage system. The storage material (solidmaterial) absorbs heat from the HTF in heat exchangers, while thesolid material and the HTF are in thermal contact.

The thermal storage capacity can be varied in order to meetdifferent load requirements, and different options are possible,depending on the storage capacity included, i.e., (i) with a smallstorage only, if electricity is only produced when the sunshine isavailable; (ii) in a delayed intermediate load conguration, wheresolar energy is collected during daytime, but with an extendedelectricity production, or a production only when demand peaks;(iii) in a fully continuous mode, with a sufciently large storagecapacity to cover electricity production between sunset andsunrise (e.g., Gemasolar).

In order to select optimum sensible heat storage materials, theheat capacity plays a major role [11,21], and values are illustratedin Fig. 7.

Molten single salts tend to be expensive [11],as illustrated inFig. 8.

The molten nitrate salt, used as HTF and storage medium, is acombination of 60 wt% sodium nitrate (NaNO3) and 40 wt%potassium nitrate (KNO3). It is a stable mixture and has a lowvapor pressure. It can be used within a temperature range of260 1C to 621 1C. However, as the temperature decreases, itstarts to crystallize at 238 1C and solidies at 221 1C [21].

4.2. Backup systems

CSP plants, with or without storage, are commonly equippedwith a fuel backup system (BS), that helps to regulate productionFig. 9. Possible combination of hybridization (a) and sole TES (b) in a solar plant.

irradiation is needed to provide a more accurate input into the

registered measurements, as discussed in Section 6, and Ho,av isthe monthly average extraterrestrial irradiation. Ho is computed

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481474for each day and location by the following formula:

Ho 24 60=pGSCdr coscosdcoswswssinsind 2With

2CSP design. It is therefore necessary to apply a methodology thatconverts these values into hourly databases. Considering that aCSP plant will only accept direct normal irradiance (DNI) in orderto operate, a clear day model is required for calculating theappropriate irradiation data.

Although numerous researchers (o2000) have generatedcalculation procedures for obtaining synthetic data on a daily orhourly basis [7,2232], the present paper updates and combinesthe essentials of these different publications into expressions ofdaily distributions and hourly variations for any selected location,starting from the monthly average solar irradiation value, bygenerating a sequence of daily and hourly solar irradiation values.Such a sequence must represent the trend of solar irradiation in aspecic area, with respect to the values observed, the monthlyaverage value and its distribution (the good and bad days).

The essential parameter is a dimensionless clearness indexvariable, dened as the ratio of the horizontal global solarirradiation and the horizontal global extra-terrestrial solar irra-diation, dened as a monthly, a daily, and an hourlycharacteristic.

In general, the meteorological variable solar radiation isneither completely random, nor completely deterministic. Highlyrandom for short periods of time (days, hours), it is deterministicfor longer periods of time (months, years). The extra-terrestrialsolar irradiation can be predicted accurately for any place andtime, since the specic atmospheric conditions of a given area willdetermine the random characteristics of the solar irradiation atground level.

5.2. The adopted model approach and equations

5.2.1. Estimating the daily irradiation

Before obtaining hourly data, estimations of daily irradiationmust be calculated rst, as shown below.

First, it is necessary to compute the monthly average clearnessindex for each month and location, which is dened as:

KT ,av Hav=Ho,av 1Where Hav is the monthly average irradiation, obtained from thewith the steam cycle and turbine already in place, only compo-nents specic to CSP require additional investment.

Fig. 9 shows a typical performance for a CSP plant enhancedwith thermal energy storage system and backup system, in aconstant generation at nominal capacity.

5. Computing global and diffuse solar hourly irradiation

5.1. Background information

To determine the optimum design and operation of the CSPthroughout the year, whilst additionally dening the potential ofTES and required BS, an accurate estimation of the daily solarirradiation is needed. Solar irradiation data for worldwide loca-tions are mostly only available as monthly averages (see Section6), and a predictive conversion into hourly data and directHo the extra-terrestrial radiation (MJ/m day)Gsc the solar constant1367 W/m2, as energy of the sun perunit time received on a unit area of the surface perpen-dicular to the propagation direction of the radiation, atmean earth-sun distance, outside of the atmosphere.

dr the inverse relative distance EarthSun, as denedbelow in Eq. (3)

ws the sunset hour angle, as dened in Eq. (4) [10]d the solar declination angle, as dened by Eq. (5) the latitude of the location (rad)n the nth day of the year (1365)

dr 10:033cos2pn=365 3The sunset hour angle, when the incidence angle is 901, as isneeded for CSP plants [33], is dened as:

cosws tantand 4The declination angle is dened by the equation of Cooper [34] as:

d 23:45sin2p284n=365 5As a result, the daily extra-terrestrial irradiation can be expressedby Eq. (6)

Ho 24 60=pGSCdrcoscosdcoswswssinsind 6Liu and Jordan [4] studied the statistical characteristics of solar

irradiation, using the clearness index (a measure of the atmo-spheric transmittance) as a random variable. They demonstratedthat the hourly clearness index was related to the monthlyaverage value. Bendt et al. [5] thereafter proposed a frequencydistribution of daily clearness index values, staring from monthlyaverage values. Initially based upon irradiation studies in the USA,this approach has been validated for different worldwide loca-tions [3336].

The distribution to the frequency of days with a value of theclearness index KT has an exponential correlation throughout themonth ranging between the minimum and maximum valuesrecorded.

The correlation is expressed as:

fKT egKT,min2egKT

h i= egKT ,min2e

gKT ,max

h i7

Where g is a dimensionless parameter that denes the particularexponential distribution, given by:

g1:4981:184x227:182e1:5x=KT ,maxKT ,min 8Where x is also a dimensionless parameter given by:

x KT ,maxKT ,min=KT ,maxKT ,av 9The minimum and maximum values of KT KT,max and KT,minrespectively, are given by:

KT ,min 0:05 10

KT ,max 0:63130:267KT,av11:9KT ,av0:758 11To obtain a daily clearness index, Knight et al. [7] dene daily KTas a function of ndk the day of the month and ndm as the numberof days of the month, with (ndk 0.5 )/ndma:

KT 1=g ln 12aegKT,minaegKT,,max

n oh i12

Finally, the daily total irradiation, H, is obtained following Eqs.(1) and (12), where the daily clearness index is multiplied withdaily extra-terrestrial irradiation H0.

H KT :H0 13In summary, with all mentioned equations solved, articial

months with articial daily total radiations (H) are created, where

months are ordered from the lowest to highest radiation level.

5.2.2. Sequence of days

Daily total radiation data results from Eqs. (1) and (13) areobtained in a predened sequence through the month by varyingradiation levels in an ascending and descending pattern;. How-ever, the sequence of days in which they succeed each other isunknown, and obviously does not strictly follow an ascending or

I0=H0, where is I0 the hourly extra-terrestrial irradiation, the

As a result, hourly global and beam irradiation data for everyday of the year (typical year of 365 days) are obtained for eachlocation, which will be used as an input for the heliostat eld.

The results of the calculations will be given and discussed inSection 7.

5.2.4. Shortcut estimates, based on recorded temperatures

The previous methodology related the radiation ux to thesunshine duration. A considerable amount of information is todayavailable on the relationship between the solar irradiation andother meteorological parameters such as cloud-cover, amount ofrain, humidity and/or temperature. The parameter that has thelargest measurement network is the ambient temperature, and ashortcut method to relate the extra-terrestrial solar irradiation tothe average daily solar irradiation.

These different methods were reviewed by Gajo et al. [40],relating H to Tmax, Tmin or Tmean.

The authors found that the original Hargreaves method per-formed overall best for different locations.

The Hargreaves method predicts KT as:

KT kRSTmax2Tmin0:5 20The adjustment coefcient kRS is empirical and differs for

interior or coastal regions:

for interior locations, where land mass dominates and airmasses are not strongly inuenced by a large water body,

reg

gh

-20

-20

-25

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481 475hourly diffuse irradiation Id is obtained as the ratio of hourlydiffuse to daily diffuse irradiation rd, which is dened as:

rd Id=Hd p=24fcoswcosws=sinwspwscosws=180g19

Finally, hourly beam irradiation Ib is calculated by subtracting Idfrom I.

Table 6Sequence model of the daily clearness indexes.

Mean clearness index (KT,av) Sequence of days throu

KT,avo0.45 24-28-11-19-18-3-2-4-90.45oKT,avo0.55 24-27-11-19-18-3-2-4-9K 40.55 24-27-11-4-18-3-2-19-9descending order, but rather present a random occurrencesequence.

Knight et al.[7] and Graham et al. [37,38] apply a separatemethodology to obtain the 31 clearness indexes which succeedeach other in a month (with 31 days) and propose a particularsequence to organize the clearness indexes as shown in Table 6.This technique is currently used to generate typical years insimulation programs such as TRNSYS [23].

5.2.3. Estimation of the hourly diffuse and beam radiation

As CSP plants accept only DNI, diffuse irradiation is subtractedfrom the global irradiation to obtain the beam irradiation, whichis the one we are interested. Direct irradiation follows a constantdirect direction, whilst diffuse irradiation is the part of the globalirradiation that follows different directions due to interactionswith the atmosphere (See Fig. 10).

The daily diffuse irradiation (Hd ) is dened by the Erbscorrelations [39]: the daily total diffuse fraction depends on thesunset hour angle (ws ) and is dened as:For wsr81.41

Hd=H 120:2727KT2:4495K2T11:951K3T9:3879K4T if KTo0:715 0:143 if KTZ0:715

14For wsZ81.41

Hd=H 10:2832KT2:5557K2T0:8448K3T ifKTo0:715 0:175 ifKTZ0:715

15

With H and Hd calculated for each day.The hourly irradiation (I) is obtained by the ratio of hourly to

daily total irradiation (rt) which is dened by the followingequation from CollaresPereira and Rabl [39] as function of thehour angle (w in radians) and the sunset hour angle (wS):

rt I=H p=24abcoswfcoswcosws=sinwspwscosws=180g16

With a and b constants given by:

a 0:4090:5016sinws60p=180 17

b 0:66090:4767sinws60p=180 18Based on Liu and Jordan [4], assuming that Id=Hd is the same asT,avFig. 10. Direct and diffuse irradiation.

the month

-14-23-8-16-21-26-15-10-22-17-5-1-6-29-12-7-31-30-27-13-25

-14-23-8-16-21-7-22-10-28-6-5-1-26-29-12-17-31-30-15-13-25

-14-23-8-16-21-26-22-10-15-17-5-1-6-29-12-7-31-20-28-13-30ional station, either because homogeneous climate conditions

tionkRS0.16; for coastal locations, situated on or adjacent to the coast of a

large land mass and where air masses are inuenced by anearby water body, kRS0.19.

The temperature difference method is recommended for loca-s where it is not appropriate to import radiation data from a

Ha(kW

4.9

3.8

3.4

7.8

6.1

7.2

6.1

7.3

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481476Table 7Selected locations with basic data.

Location Latitude

(rad)

Longitude

(rad)

Hav (Jan)

(kW h/m2 day)

Chuquicamata, Chile 22.5 68.9 8.32Upington, RSA 28.5 21.08 7.93Geraldton, Australia 28.78 114.61 8.28Sevilla, Spain 37.41 5.98 2.56Font Romeu, France 42.5 2.03 1.81

Marrakech, Morocco 31.6 8 3.49Sainshand, Mongolia 44.89 110.14 2.21

Ely, USA 39.3 114.85 2.56

a Data taken from nearby Albi.b Data taken from nearby Ulan Bator.do not occur, or because data for the region are lacking. For islandconditions, the methodology of Eq. (20) is not appropriate due tomoderating effects of the surrounding water body.

Since Tmax and Tmin data are indeed widely available, theHargreaves KT-values can be used in the methodology of Section5.1, and results of the both methods will be illustrated in Section 7.

6. Model parameters

6.1. Common measurement methods of solar radiation

Solar radiation can be measured with pyranometers, radio-meters or solarimeters. The instruments contain a sensor installedon a horizontal surface that measures the intensity of the totalsolar radiation, i.e., both direct and diffuse radiation from cloudyconditions. The sensor is often protected and kept in a dryatmosphere by a glass dome that should be regularly wipedclean. Where pyranometers are not available, solar radiation isusually estimated from the duration of bright sunshine. Theactual duration of sunshine, n, is measured with a CampbellStokes sunshine recorder. This instrument records periods ofbright sunshine by using a glass globe that acts as a lens. Thesun rays are concentrated at a focal point that burns a hole in aspecially treated card mounted concentrically with the sphere.The movement of the sun changes the focal point throughout theday and a trace is drawn on the card. If the sun is obscured, the

Fig. 11. Calculated values of H0.v (Jul)

h/m2 day)

Jan Jul

Tmax (1C) Tmin (1C) Tmax (1C) Tmin (1C)

9 25.2 5.8 21.1 1.5

9 36.7 20.5 21.5 1.5

1 33.6 19.3 19.7 9.1

0 15.9 7.6 35.7 19.9

7 9.5a 0.1a 26.9a 14.2a

6 21.4 6.6 37.3 19.8

1 18.5b 34.7b 23.5b 9.4b8 3.3 13 31.3 9.7trace is interrupted. The hours of bright sunshine are indicated bythe lengths of the line segments.

6.2. Available information

There are two reliable sources that provide information on thetwo of the most basic meteorological parameters: monthly meantemperature and solar radiation. These sources are the NASAwebsite [41] and TUTIEMPO [42]. NASA has produced a grid mapof the longitude. The solar radiation data are an estimate that hasbeen produced from satellite-based scans of terrestrial cloud-cover. Note that NASA does not provide the mean-daily maximumand minimum temperature. TUTIEMPO on the other hand pro-vides daily mean, maximum and minimum temperature data forany given location. The data are based on measurements carriedout by a wide network of meteorological stations and hence theselatter data are very reliable. Note that the NASA data are availableon a mean-monthly basis, whereas TUTIEMPO are downloadableon a day-by-day basis. It is important to remember that NASAdata are based on satellite observations that represent inferredvalues of irradiation; in contrast, TUTIEMPO provides ground-measured data for temperature. Hence, if reliable regressions areavailable between irradiation and mean temperature, then thelatter data may be used to obtain more realistic estimates ofirradiation.

Fig. 12. Calculated average monthly clearness index by the model [Eq. (1)]and by the Hargreaves method [Eq. (21)] at Upington (RSA).

n in (a), January (summer) and (b), July (winter).

Fig. 14. The monthly average Id/I ratio for Chuquicamata (Chile).

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481 477Having developed the underlying equations of the calculationmethod in Section 5, and using the data acquisition of Section 6,the present section will illustrate the use of the data obtained.

The monthly extra-terrestrial irradiation, H0, computed by Eq.(6), is illustrated in Fig. 11 for different latitudes in both hemi-spheres and shows the seasonal dependence, whilst also illustrat-ing the maximum and minimum values obtained throughout6.3. Selected locations

To illustrate the use of the methodology of Sections 5, 8locations were selected because of being already associated withCSP, or announced as potential sites for STP. The essential data ofthe locations are given is Table 7.

7. Results and discussion

7.1. Calculations of H0, H and Hb

Fig. 13. The total daily radiation for Upingtothe year.To proceed with the calculation of the monthly average

clearness index, KT,av Eq. (1) is used together with NASA-data[41]. Results are illustrated as example in Fig. 12 for the Upingtonlocation. The Figure also includes the results obtained from theHargreaves method [Eq. (21)] using TUTIEMPO-data [42]. Clearly,both methods provide similar results of KT,av for most of themonths, however with higher Hargreaves-values in Spring andAutumn. The model approach thus provides slightly more con-servative KT,av values, and is recommended for design.

The daily total irradiation is thereafter obtained by applyingthe daily clearness index, KT, and the daily extra-terrestrialirradiation H0. Fig. 13 shows the model-predicted total dailyirradiation, ordered in ascending daily pattern for Upington, fora summer month (January) and a winter month (July). Themonthly average H, calculated by Eq. (13) in January and July, is7.92 kW h/m2-day and 3.92 kW h/m2-day, respectively. Similarevolutions can be obtained for the other selected locations.Applying the Knight et al. [7] sequence model for the dailyclearness indexes, as function of KT,av transforms the ascendingnature of the consecutive days into a wave-function, althoughmonthly average values of H remain unchanged.

A similar evolution can be predicted using daily KT-valuesresulting from the daily Tmax and Tmin data, according to Eq. (21).

Fig. 15. The daily solar beam irradiation, Hb, on the 15th of January and onthe 15th of July , for different locations in both the Southern and Northern

Hemisphere.

These results are also shown in Fig. 13. Clearly, the Hargreavesapproach provides a more constant H-value throughout themonth since not respecting an ascending daily pattern. Themonthly average value of H (HRG) is closely related to the modelpredictions: 7.56 kW h/m2-day in January, and 4.07 kW h/m2-dayin July, i.e., a deviation of 4.5% and 3.8% only with the pre-citedvalues of the model-predicted average values.

The most important result towards CSP design requires thedirect (beam) irradiation, obtained by withdrawing the diffuseirradiation, Hd, from the total irradiation, H, according to Eqs. (14)and (15).

The ratio of the diffuse to total irradiation is illustrated forChuquicamata in Fig. 14: the more cloudy winter season (AprilAugust) is reected in the higher value of the ratio.

The resulting beam radiation, Hb, for representative days inSummer and winter, for all locations, is shown in Fig. 15, whereasa more detailed monthly average evolution for some locations, isshown in Fig. 16.

Finally, a complete hourly evolution can be predicted by the

The efciencies of the essential components has been reportedby S&L, and represented in Table 9.

DNI is calculated on hourly bases The total energy ux reected by the heliostat eld is

calculated The expected nominal capacity of the plant is selected 21 consecutive days of lowest radiation levels are selected to

coincide with the maintenance period, thus limiting lossesduring plant stand-still

From a given starting day of the year, e.g., January 1st., at 6:00a.m., and repeated for all hours of the year, the followingdifferent options need to be assessed:

If the solar thermal ux exceeds the required value tooperate the plant at nominal capacity, only solar thermalenergy will be used, whilst excess solar energy is stored inthe HTF hot storage tank. The BS-system is not used, andadditional excess solar thermal energy cannot be used;

If the solar thermal ux is insufcient to meet the nominalcapacity, but enough thermal energy is stored in the hottank, no BS is needed, and the plant will operate oncombined solar radiation and stored energy;

If the combined solar thermal ux and energy stored areinsufcient, the plant needs to operate in its hybrid con-guration, using the BS to meet the thermal requirements.

The detailed simulations are extensive, and are not included in

Pro

nnual overall CSP efciency (%) 13.0 13.7 16.1 16.6 17.3 18.1

Source of estimation S&L SNL S&L SNL S&L SNL

Table 9Values of CSP-component efciencies.

Component Efciency (%)

Solar eld 4850

TES 499Power block 40

Fig(Ch

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481478Considering that about 10% of the generated electricity will beused internally for the plant utilities (mostly pumping), 90% of thecombined efciencies do indeed vary between 14 and 18%.

The nal CSP performance simulation follows the strategy ofFig. 18, with a specic algorithm to be used, in terms of DNI, TESand BS, as previously presented in Fig. 9.model, as illustrated in Fig. 17, where the radiation ux can beseen to increase from sunrise to noon, and thereafter decreasingagain till sunset.

It is also clear that the selection of the CSP nominal capacitywill be a compromise between the seasons, accounting for thecapability of thermal storage, and the use of a backup system.

7.2. Methodology to apply the predictions in CSP design

Having established the annual, monthly and daily levels ofdirect (beam) solar irradiation, its impact on the power yield ofthe CSP can be assessed. To do so, it should be remembered thateach of the operations of the overall CSP-layout has its ownefciency, reected in its overall efciency. The projected overallefciency of CSP plants was assessed by S&L and SNL, aspresented in Table 8, including projected increased efcienciesas a result of present and future improvements.Fig. 16. Evolution of the average monthly direct (beam) irradiation in 8 locations.Ye

Athear of projection 2004 2004 2008 2008 2020 2020le 8jected overall CSP efciency.Tab. 17. Hourly evolution at the 15th of the respective months, in Chuquicamataile).present paper. They will be reported upon in a follow-up

stri

ce si

orage anks

Steam generation Turbine Grid

10%

e components of the SPT.

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481 479paper, considering the application of the solar tower collector,with molten salt HTF/TES, and with natural gas-red BS. Theturbine capacity will be 19.9 MWel, chosen because of theextensive data available for the Gemasolar plant (Spain). Suchsimulations will be carried out for those locations where theannual average daily irradiation ux exceeds4 kW h/m2-day.

Due to parasitics (electricity use within the plant for pumps,cooling towers, compressors, instrumentation and controls, light-ing, heat tracing,.), estimated at 10% for a Gemasolar-type hybridapplication, the net output to the grid will be 17.9 MWel during theperiod of operation. The annual power yield of the hybrid Gema-solar plant is 110 GW h/year [2]. The specic energy production,i.e., the ratio of total annual grid energy and the rated net power

Heliostats Field Characteristics

Solar Radiation Data

Thermal Energy Storage Characteristics

Plant Efficiencies

Plant Re

Fig. 18. Performan

Heliostat field Receiver

Thermal field

circuitStt

Parasitics (

Fig. 19. Sequence of thoutput of the plant is therefore 110,000/17.96145 h/year. Fortotal of 8760 h/year, the overall yield is 70.1%, assumed realistic inview of the annual maintenance period (zero production), shortduration disturbances, and very low solar irradiation in winter. Aswill be shown hereafter for the Chiquicamata example, 69.5 GW h/year will be generated from the solar energy alone (the remainingturbine power being generated by the back-up natural gas system).The average net solar yield thus represents 69,500/17.93882 h/year, or 44.3% (for the 8760 h annual operation).

For the simulations, it is considered that the plant works witha thermodynamic cycle in a steady state.

If a total energy production needed from the turbine is19.9 MWel, the efciencies of the different CSP components willdetermine the hourly heat ow along the system, from heliostatsto turbine and grid. The system ow sheet is illustrated in Fig. 19,and the additional required information for each step of thesequence is given below in different notes.

The total energy of 19.9 MWel in the steam turbine must bereached each hour, with part of this energy added to the moltensalts in the receiver (reected by the heliostat eld), and the restof the energy added by the storage system, and/or by the back-upsystem (when receiver and storage energies are insufcient)according to Fig. 9 and the strategy of Fig. 18. By simulating theplant performance at the heliostat and receiver levels, the addi-tional energy required by the storage system and the backupsystem can be determined. Molten salt properties at each point ofthe thermodynamic cycle are xed and knownctions

Hourly Stored Energy

Demand Supplied by Solar Energy

Short and Long Term Backup Requirements

mulation method.

)Note 1. Heliostat eld

Since the heliostats follow the sun by a two-axis tracking, nocorrection for the incident angle ymust be made (cos y1), and Ibcorresponds to the real hourly irradiation at the heliostat eld.

First, the energy reected by the heliostat eld must be calculatedeach hour, where hourly radiation data is extracted from thecalculations of Sections 5 and 6 before. The total energy reectedby the heliostat eld and concentrated in the receiver is thendetermined by the heliostat eld efciency and the heliostat eldreective surface area. The heliostat eld efciency (ZHF ) is mostlycharacterized by its reectivity, optical efciency, heliostat corrosionavoidance and cleanliness. A value of 48 to 50% is commonly used(Table 9).

Note 2. Receiver

The reected energy is concentrated in the receiver, which actsas a heat exchanger where circulating molten salts absorbs solarenergy. Total energy absorbed by molten salts is determined bythe receiver efciency, where receiver thermal losses are primar-ily driven by the thermal emissivity of the receiver panels

Fig. 20. Electricity generation through the year in a Chuquicamata SPT, only withsolar resource conguration.

e Ch

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481480(radiation losses) and by the receiver temperature. Commonly,radiation losses are 51% [14].

Note 3. THF Circuit and THF storage

Piping and tank losses are very limited, due to the efcientisolation applied, normally again o1% [14].

Note 4. Steam generation and reheated Rankine cycle

The efciencies to be considered include the design point turbinecycle efciency; start-up losses; partial load operation, and steamgeneration system efciency. Losses due to minimum turbine loadrequirements do not apply since the plant has a thermal storage andback-up system. Commonly, a value of 40% is assigned to the overallefciency of the thermal power block of the plant (Table 9).

Note 5. Parasitics (in-plant energy use)

The parasitic consumption considers internal electricity usagemostly in heliostat tracking, THF pumps, condensate pumps,

Fig. 21. Monthly backup requirements in thfeedwater pumps, cooling water pumps, cooling tower fans andelectrical heat tracing system, but additionally in instrumenta-tion, controls, computers, valve actuators, air compressors, andlighting. A maximum 10% [13] was measured by Sargent andLundy Consulting Group

Considering component efciencies, a simple estimation ofsolar to electricity efciency (Zsolar) of the plant can be obtainedby using component efciencies to calculate the total efciency ofenergy transformation.

Zsolar ZHF Zrec Zpiping Zstorage Zcycle 1Zparasitic A

Where,

ZHF: Heliostat eld efciency,Zrec: Receiver efciency,Zpiping: Piping efciency,Zstorage: Thermal storage efciency,Zcycle: Power block gross efciency,Zparasitic: Parasitics,A Plant availability (capacity factor).

Initial results of the simulation for the Chuquicamatainitiative, reveal that the solar generation will account for69.5 GW h/year, in the case of a conservative 13% overallefciency.

Figs. 20 and 21 provide some indications of the simulatedresults. The zero production between days 150 and 180 corre-spond with the supposed annual shut-down period for overallmaintenance.

Provided overall efciencies will increase over the comingyears, due to technical improvements, the solar energy contribu-tion will increase, thus reducing the required backup, as will bediscussed in a follow-up paper, considering the overall design ofthe SPT plant.

8. Conclusions

To determine the optimum design and operation of the CSPthroughout the year, whilst dening the required TES and/or BS,an accurate estimation of the direct daily solar irradiationis needed

The paper develops the underlying equations to calculate the

uquicamata SPT, in hybrid operating mode.monthly extra-terrestrial irradiation, H0,the monthly average clearness index, KT,av the daily total

irradiation, and the direct (beam) irradiation.Results of the model approach is given for 8 selected locations,

in both Northern and Southern hemisphere.Having established the annual, monthly and daily levels of

direct (beam) solar irradiation, its impact on the power yield ofthe CSP can be assessed. The projected overall efciency of CSPplants was assessed and included in a CSP performance simula-tion, according to a proposed strategy. Initial simulation resultsare illustrated for a 19.9 MWel Solar Power Tower project, withmolten salts as HTF, and operating in an hybrid way (includingheat storage and back up fuel). In the assesses example, solargeneration will account for 69.5 GW h/year, in the case of aconservative 13% overall efciency. Provided overall efciencieswill increase over the coming years, due to technical improve-ments, the solar energy contribution will increase, thus reducingthe required backup, as will be discussed in a follow-up paper,considering the overall design of the SPT plant.

References

[1] /http://www.alternative-energy-resources.net/solarenergydisadvantages.htmlS.

[2] /http://www.torresolenergy.com/TORRESOL/gemasolar-plant/enS.[3] Dufe J, Beckmann W. Solar engineering of thermal processes. 3rd ed.. USA:

John Wiley and Sons; 2006.[4] Liu BYH, Jordan RC. The interrelationship and characteristic distribution of

direct, diffuse and total solar radiation. Solar Energy 1960;4:119.[5] Bendt P, Collares-Pereira M, Rabl A. The frequency distribution of daily

insolation values. Solar Energy 1981;27:15.[6] Stuart R, Hollands K. A probability density function for the beam transmit-

tance. Solar Energy 1988;40:4637.[7] Knight K, Klein K, Dufe J. A methodology for the synthesis of hourly weather

data. Solar Energy 1991;46:10920.[8] OECD/IEA, technology roadmap, concentrating solar power, 2010.[9] Muller-Steinhagen H, Trieb F. Concentrating solar power: a review of the

technology. Ingenia 2004;18:4350.[10] Llorente I, Alvarez JL, Blanco D. Performance model for parabolic trough solar

thermal power with thermal storage: comparison to operating plant data.Solar Energy 2011;85:244360.

[11] Barlev D, Vidu R, Stroeve P. Innovation in concentrated solar power. SolarEnergy Materials & Solar Cells 2011;95(10):270325.

[12] SolarPACES, annual Reports, 2006, 2007, 2008, 2009, 2010. Downloadablefrom: /http://www.solarpaces.org/Library/AnnualReports/annualreports.htmS.

[13] Sargent and Lundy Consulting Group, assessment of parabolic trough andpower tower solar technology cost and performance forecasts, NationalRenewable Energy Laboratory, 2003.

[14] Ortega J, Burgaleta J, Tellex F. Central receiver system solar power plant usingmolten salt as heat transfer uid, May 2008.

[15] Pitz-Paal R, Dersch J, Milow B, Tellez F, Ferriere A, Langnickel U., et al.Development steps for concentrating solar power technologies with max-imum impact on cost reduction, 2005. Available at: /http://pre.ethz.ch/

[24] Pissimanis D, Karras G, Notaridou V, Gavra K. The generation of a typicalmeteorological year for the city of Athens. Solar Energy 2000;40:40511.

[25] Aguiar RJ, Collares-Pereira M, Conde JP. Simple procedure for generatingsequences of daily radiation values using a library of Markov transitionmatrices. Solar Energy 1988;40:26979.

[26] Aguiar RJ, Collares-Pereira MTAG. a time-dependent, autoregressive, Gaus-sian model for generating synthetic hourly radiation. Solar Energy1992;46:16774.

[27] Amato U, Andretta A, Bartoli B, Coluzzi B, Cuomo V. Markow Processes andFourier analysis as a tool to describe and simulate daily solar irradiance. SolarEnergy 1986;37:17994.

[28] Graham VA, Hollands KGT, Unny TE. A time series model for K, withapplication to global synthetic weather generation. Solar Energy 1988;40:8392.

[29] Graham VA, Hollands KGT. A method to generate synthetic hourly solarradiation globally. Solar Energy 1990;44:33341.

[30] Gordon JM, Reddy TA. Time series analysis of daily horizontal solar radiation.Solar Energy 1988;41:21526.

[31] Mora-Lopez LL, Sodracj de Cardona M. Characterization and simulation ofhourly exposure series of global radiation. Solar Energy 1997;60:25770.

[32] Mora-Lopez LL, Sodracj de Cardona M. Multiplicative ARMA models togenerate hourly series of global irradiation. Solar Energy 1998;63:28391.

[33] Tham Y, Muneer T, Davison B. Estimation of hourly averaged solar irradia-tion: evaluation of models. Building Services Engineering Research andTechnology 2010;31(1):925.

[34] Muneer T, Fairooz F. Quality control of solar radiation and sunshinemeasurements-lessons learnt from processing worldwide database. BuildingServices Engineering Research and Technology 2002;23(3):15166.

[35] Hawas MM, Muneer T. Study of diffuse and global radiation characteristics inIndia. Energy Conversion and Management 1984;24(2):1439.

[36] Badescu V, Gueymard CA, Cheval S, Oprea C, Baciu M, Dumitrescu A, et al.Computing global and diffuse solar hourly irradiation on clear sky. Reviewand testing of 54 models. Renewable and Sustainable Energy Reviews

H.L. Zhang et al. / Renewable and Sustainable Energy Reviews 22 (2013) 466481 481sun.net/book.htmlS.[18] Abengoa solar website. /http://www.abengoa.esS.[19] N.R.E.L. Website, with the collaboration of SolarPaces, /http://www.nrel.gov/

csp/solarpaces/S.[20] /www.guardian.co.uk/environment/2010/oct25/

south-africa-solar-power-plantS.[21] Fernandes D, Pitie F, Caceres G, Baeyens J. Thermal energy storageHow

previous ndings determine current research priorities. Energy 2012;39(1):24657.

[22] Klein SA, Beckman WA. TRNSYS V14.2 (1996). A transient system simulationtool.

[23] Petrakis M, Kambezidis HD, Lykoudis S, Adamopoulos AD, Kassomenos P,Michaelides IM, et al. Generation of a typical meteorological year for NicosiaCyprus. Renewable Energy 1998;13(3):3818.application to global synthetic weather generation. Solar Energy 1988;40(2):8392.

[38] Graham VA, KGT. Hollands. A method to generate synthetic hourly solarradiation globally. Solar Energy 1990;44(6):33341.

[39] Tham Y, Muneer T, Davidson B. Estimation of hourly averaged solar irradia-tion: evluation of models. Building Service Engineering Research and Tech-nology 2010;31(1):925.

[40] Gajo EJ, Etxebarria S, Tham Y, Aldali Y, Muneer T. Inter-relationship betweenmean-daily irradiation and temperature, and decomposition models forhourly irradiation and temperature. International Journal of Low-carbonTechnologies 2011;6:2237.

[41] /http://eosweb.larc.nasa.gov/cgi-bin/sse/retscreen.cgi?email=rets@nrcan.gc.caS.[42] /http://www.tutiempo.net/en/S.[17] Stine W, Geyer M. Power from the Sun, 2001, /http://www.powerfromthe 2012;16:163656.[37] Graham VA, Hollands KGT, Unny T. A time series model for Kt withpublications/journals/full/j138.pdfS.[16] Pitz-Paal R, Dersch J, Milow B, Romero M, Tellez F, Ferriere A., et al. European

Concentrated Solar Thermal Road-Mapping-Executive Summary, CEECOSTARContract: SES6-CT-2003-502578. (2005) 144.

Concentrated solar power plants: Review and design methodologyIntroductionSolar irradiance as worldwide energy sourceConcentrated solar power plants

CSP technologiesGeneralitiesSolar power towersParabolic trough collectorLinear Fresnel reflectorParabolic dish systemsConcentrated solar thermo-electrics

Comparison of CSP technologies

Past and current SPT developmentsEnhancing the CSP potentialThermal energy storage systemsBackup systems

Computing global and diffuse solar hourly irradiationBackground informationThe adopted model approach and equationsEstimating the daily irradiationSequence of daysEstimation of the hourly diffuse and beam radiationShortcut estimates, based on recorded temperatures

Model parametersCommon measurement methods of solar radiationAvailable informationSelected locations

Results and discussionCalculations of H0, H and HbMethodology to apply the predictions in CSP design

ConclusionsReferences