Concentrated solar p

Alexander H. Slocum a,, Daniel S. Codd a, Jacopo Buongiorno b, Charles Forsberg b,b c d a

result, high temperature thermal energy can be provided 24/7 or at any desired time.

thermal systems that enable energy storage with electricityproduction when there is limited sunlight. Concentratingsolar power (CSP) systems that use a central receiver withintegral thermal energy storage have the potential to pro-duce 24/7 base load and/or peak electric power. Powertowers use heliostats to focus sunlight on a receiver placed

617 258 6427.E-mail addresses: slocum@mit.edu (A.H. Slocum), codd@mit.edu

(D.S. Codd), jacopo@mit.edu (J. Buongiorno), cforsber@mit.edu (C.Forsberg), tmckrell@mit.edu (T. McKrell), jcnave@math.mcgill.ca (J.-C.Nave), cnp@cyi.ac.cy (C.N. Papanicolas), ghobeity@mit.edu (A. Ghobe-ity), noone@mit.edu (C.J. Noone), stefanop@mit.edu (S. Passerini),folkersr@mit.edu (F. Rojas), amitsos@alum.mit.edu (A. Mitsos).

Available online at www.sciencedirect.com

Solar Energy 85 (2011)The amount of storage required depends on local needs and economic conditions. About 2500 m3 of nitrate salt is needed to operate a4 MWe steam turbine 24/7 (7 h sunshine, 17 h storage), and with modest heliostat eld oversizing to accumulate energy, the system couldoperate for an additional 24 h (1 cloudy day). Alternatively, this same storage volume can supply a 50 MWe turbine for 3.25 h withoutadditional solar input. Cosine eect losses associated with hillside heliostats beaming light downwards to the receiver are oset by theelimination of a tower and separate hot and cold storage tanks and their associated pumping systems. Reduced system complexity alsoreduces variable costs. Using the NREL Solar Advisor program, the system is estimated to realize cost-competitive levelized productioncosts of electricity. 2011 Elsevier Ltd. All rights reserved.

Keywords: Concentrating solar power; Molten salt; Volumetric absorption receiver; Hillside heliostats; Thermal storage

1. Background

A robust solar energy portfolio is likely to include solar

Corresponding author. Address: Department of Mechanical Engineer-ing, Massachusetts Institute of Technology, 77 Massachusetts Ave.#3-445, Cambridge, MA 02139, USA. Tel.: +1 617 253 0012; fax: +1Thomas McKrell , Jean-Christophe Nave , Costas N. Papanicolas , Amin Ghobeity ,Corey J. Noone a, Stefano Passerini b, Folkers Rojas a, Alexander Mitsos a

aDepartment of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USAbDepartment of Nuclear Science & Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA

cThe Department of Mathematics and Statistics, McGill University, Montreal, Canada H3A 2K6dThe Cyprus Institute, 15 Kypranoros Street, Nicosia 1061, Cyprus

Received 22 November 2010; received in revised form 21 March 2011; accepted 11 April 2011Available online 11 May 2011

Communicated by: Associated Editor Robert Pitz-Paal

Abstract

A concentrating solar power system is presented which uses hillside mounted heliostats to direct sunlight into a volumetric absorptionmolten salt receiver with integral storage. The concentrated sunlight penetrates and is absorbed by molten salt in the receiver through adepth of 45 m, making the system insensitive to the passage of clouds. The receiver volume also acts as the thermal storage volumeeliminating the need for secondary hot and cold salt storage tanks. A small aperture and refractory-lined domed roof reduce lossesto the environment and reect thermal radiation back into the pond. Hot salt is pumped from the top of the tank through a steam gen-erator and then returned to the bottom of the tank. An insulated barrier plate is positioned within the tank to provide a physical andthermal barrier between the thermally stratied layers, maintaining hot and cold salt volumes required for continuous operation. As a0038-092X/$ - see front matter 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.solener.2011.04.010ower on demand

www.elsevier.com/locate/solener

15191529

Eneatop a tower to reduce heliostat shadowing, increase opti-cal eciency, and to achieve high solar ux concentrationand steam plant eciency (Viebahn et al., 2008). Conven-tional high temperature CSP systems have evolved to usea central tower where a heat transfer uid circulatesthrough tubes onto which the sunlight is focused. However,maximum allowable uxes are limited to avoid thermaldegradation of the receiver tubing. Lata et al. (2008) citesthe tradeos between tube diameter, wall thickness, recei-ver durability and pressure drop in conventional tubularreceiver designs while describing an external tubular recei-ver capable of achieving slightly higher maximum uxes,up to 1.0 MW/m2, thereby reducing receiver surface areaand losses while increasing overall plant eciency.

Utilizing molten salts as the working uid enables sim-ple subsequent thermal storage, due to their high heatcapacities and wide operating temperatures. Unfortu-nately, daily receiver lling requires ancillary heaters andpresents additional risks should the salt freeze, requiringelectric heat tracing on long piping runs, valves and mani-folds. Despite these measures, operating problems stilloccur; for example, the Solar Two CSP demonstrationplant was disabled by frozen salt in pipes (Reilly and Kolb,2001). Another approach is direct absorption of sunlight byseveral-centimeter thick salt waterfalls, but the cost ofpumps, manifold and piping preheaters, and uid owvariations as a function of varying solar ux, limited the

Nomenclature

Abbreviations

CSPonD Concentrated Solar Power on Demand

Symbols

a attenuation coecientd optical thicknesse emissivity/ beam-downanglehi incident angle

1520 A.H. Slocum et al. / Solarpracticality of such systems (Bohn, 1987).To minimize receiver uid pumping losses and enable

alternative receiver designs, Rabl (1976) proposed abeam-down reective tower with a ground-based CSPreceiver. Similarly, Yogev (1997) and Epstein et al. (1999)suggested a beam-down system where the light was to bebeamed directly into a molten salt/metal lled container.It has even been proposed for tall buildings to use bal-cony-mounted heliostats to direct sunlight to a receiverplaced atop the building (LeBarre, 2010). Since Rabls pro-posal in 1976, signicant experimental work has occurredon beam-down towers and ground receivers, especiallyfor reforming materials (Yogev et al., 1998; Segal andEpstein, 2003). The Odeillo solar furnace facility uses anorth facing parabola focused on a target built into onewall of a building that holds oces and laboratories.Sixty-three south-facing at mirror heliostats track thesuns movement and focus it down on the north facingparabola (Trombe and Le PhatVinh, 1973). NREL alsohas a high-ux solar furnace system where heliostats aimlight towards a ground-based secondary reector systemthat redirects and concentrates the sunlight to a small aper-ture receiver, although up to 10% of the energy is lost witheach reection (Skinrood et al., 1974). These systemsachieve high concentrations and receiver temperatures withlarge, precision secondary optical elements whose costhas prevented the commercial adoption of beam-downCSP.

Conventional thermal storage systems require two mol-ten salt storage tanks, each capable of storing the entiresystem volume (Herrmann et al., 2004). In traditionalCSP systems, cost-savings have been obtained with singletank systems relying on temperature stratication vianatural thermocline formation (Pacheco et al., 2002).Copeland and Green (1983) and Copeland et al. (1984)have shown rafted thermocline designs eective at boost-ing thermal stratication in water tanks, with suggesteddesigns for molten salt thermal storage tanks. However,passive rafted thermoclines would rely on two parametersdicult to control in high temperature molten salt tanks:maintaining neutral buoyancy at the hotcold thermoclineinterface; and a near perfect seal with the side walls toprevent leakage around the divider raft. Indeed, tests

ht transmitted anglehr reected angleD diameterIo incident intensityI transmitted intensityn index of refractionR reection coecient

rgy 85 (2011) 15191529performed in water showed the neutrally buoyant raftdesign may display instabilities and tilt and/or jam in thestorage tank. Demonstration CSP plants without storagehave typically been designed with a co-ring gas turbinescheme to enable continued operation when the solarsystem is down (European Commision, 2005).

2. CSPonD: collocated receiver and storage system

Here we present a new system with heliostats mountedon a hillside that beam light directly into an open containerof molten salt at the base of the hill, or into a one-bouncesystem with the receiver at the top of the hill (Fig. 1). Asmall aperture in the receiver lets the sunlight penetratethe surface of the large molten high temperature salt pond.Volumetric absorption enables a simpler receiver design

ceivthert an

Ene(A) Beam-downFig. 1. Representative CSPonD sites in White Sands, NM for pond resimultaneously collects sunlight while also acting as the beam-down opticlocation and discretized terrain sampling locations from elevation data, no

Non-imaging refractory lid

Lid heat extraction(A)

A.H. Slocum et al. / Solarwith a free surface molten salt pond, without high-pressure,high-ow molten salt pumps, capable of high temperaturestorage. The molten salt surface is self-healing toleratinghigh solar ux transients without irreparable sudden orcumulative damage to the receiver. Concentrated SolarPower on Demand (CSPonD) could provide 24/7 powerand thus help ll a critical need in solar power, that ofenergy storage (Slocum et al., 2010). Incoming concen-trated solar ux directed at the aperture can follow oneof three paths: refracted into the molten salt; reected othe salt surface towards the inner surface of the receiverlid; or directly impinged on the inner surface of the lid(Fig. 2). An important design goal is to shape the coverto function as a diuse reecting concentration booster,not unlike the compound parabolic concentrator units usedin beam down towers (Bassett and Derrick, 1981; Derrickand Bassett, 1985).

For the near term we consider a salt commonly used inCSP plants: sodiumpotassium nitrate (e.g., Hitec solarsalt: 60/40 wt.% NaNO3KNO3) which has a low meltingpoint of 222 C. Although above 593 C solar salt decom-poses and becomes corrosive and dangerous, systems havebeen built and operated to pump it between hot and cold

Hot salt to HX

Cold salt from HX

Fig. 2. Cross sections of a CSPonD volumetric absorption molten salt receiver/of a sunny day the divider plate has moved down and the hot salt region is fullythe divider plate has moved up and the cold side is full. Hot molten salt is pupower and temperature input to a steam generator or other power cycle. Coldethe tank. (For interpretation of the references to color in this gure legend, th(B) Beam-uper without (A) and with (B) secondary reector. The CSPonD systemeby reducing overall system complexity and cost. Icons illustrate receiveractual heliostat layout. See Noone et al. (2011).

Insulated aperture doors (B)

rgy 85 (2011) 15191529 1521storage tanks and a steam generator (Herrmann et al.,2004). The power block, including salt pumps, heat exchan-ger/steam generator and power generation device, for anitrate salt based CSPonD system will be very similar tothose that can be commercially obtained. Hence, for a nearterm CSPonD system, a steam power cycle will be assumedwith peak steam temperatures of 500540 C.

Both the molten salt pond and the lid will exchange heatwith each other and to the environment by radiation, con-vection and conduction; the primary heat transfer mecha-nism is through radiation. The system cover will be linedwith refractory rebrick and backside cooled, so the saltvapor rises and condenses on the inner surface of the cover,akin to frost collecting on evaporator coils within a refrig-erator. The resulting white surface will grow until the thick-ness results in a thermal resistance that condenses the saltvapor, but the surface continually melts and returns liquidsalt to the pond. The liquid/solid interface is expected toact as a diuse reector to incoming light that reects othe surface of the salt. Solidied salt has a very high reec-tivity in the visible spectrum and behaves nearly as aLambertian reector; liqueed salt will be subject to graz-ing-angle Fresnel surface reections and as such, much of

Hot salt

Cold salt Divider plate

storage system: (A) Yellow arrows denote light path within receiver. At endcharged. (B) Whereas after prolonged heat extraction without solar input,mped as needed from the top of the tank to provide reasonably constantr, yet still liquid salt from the steam generator is returned to the bottom ofe reader is referred to the web version of this article.)

for a nominal 20 cm annular clearance on the proposedsystem), accounting for the daily downward motion ofthe plate displacing uid and pumped cold salt returningfrom the heat exchanger. Semenyuk (1983) describes ameans to calculate exergy losses for mixing two workingbodies diering only in temperature. For a nitrate CSPonDreceiver with hot salt at 55 C, cold salt at 25 C and ambi-

0.0

0.2

0.4

0.6

0.8

1.0

01020304050607080

0 4 8 12 16 20 24 28 32 36 40 44 48

Div

ider

pla

te p

ositi

on (z

/H)

Pow

er (M

Wt)

Time (h)

divider plateposition

extracted power

transient solar input

ideal solar input

Fig. 3. Simplied divider plate position example for two day/night cycles:Ideal solar input (left) and transient, or cloudy, solar input (right) for aconstant 20 MWt extraction system.

50

150

250

350

450

550

Tem

pera

ture

(C

)

Distance from tank exterior

Tsalt = 550 C

Tsalt = 250 C

Fig. 4. Tank wall insulation schematic. Internal insulating refractorybrick is used to reduce temperature uctuations at the tank shell. A thin,thermally conductive alloy liner can be employed to reduce axial thermal

Energy 85 (2011) 15191529the light impinged onto the lid will be reected back to thepond. The energy that is transferred to the lid is fromlower-temperature, longer wavelength radiation from thesalt surface. Only a small fraction of incoming photonenergy is converted to thermal energy at the lid.

The vapor pressures of molten salts are fairly low, on theorder of 0.001 bar for chloride salts at 900 C. As a result,the overall fuming rate is low, given as 0.2 kg/m2 ofexposed salt area per hour for a chloride salt bath at870 C (ASM, 1991). For the 5 m deep 25 m diametersystem proposed in the following text, this equates toroughly 100 kg/h, or less than 0.06% of the entire salt mass.Lower temperature nitrate CSPonD systems will havelower salt vapor pressures and mass transfer, and reducedradiative transfer between the salt surface and the lid ascompared to higher temperature chloride salt systems.The lid is not depending on mass transfer for cooling active cooling will be employed to obtain the desired lidoperating temperature. Cooling loops would be concen-trated in high heat ux regions, and various zones can beemployed for temperature control throughout the lid. Itmay be found that the optimal lid design is not isothermal,but has varying temperature to limit radiative transfer andconvective losses. The collected cover energy, unique toCSPonD systems, can vary from 2% to 20% of the incidentsolar power and this intermediate-temperature heat is usedelsewhere in the plant (Ghobeity et al., submitted for pub-lication). The percentage of lid heat extracted depends pri-marily on the plant layout: hillside topology, operatingtemperatures, and seasonal and diurnal position of thesun. In a dual-purpose desalination and electricity produc-tion plant using the CSPonD concept, heat collected by thecover can be used for preheating feed water to the steamgenerator or desalination feed water (Goosen et al., 2002;Ghobeity and Mitsos, 2009; Ghobeity et al., submittedfor publication).

The top surface of the salt needs to remain at a constantlevel for consistent solar absorption; hence, the tank is splitinto two zones with a moving barrier plate. The top zone isthe hot salt side, and the bottom zone is the cold salt side.An insulated creep and corrosion resistant alloy platedivides the two sides, providing a physical and thermal bar-rier between the thermally stratied hot and cold layerswithin the tank. The light will penetrate deeply and a smallfraction of it will impact the highly absorbing divider platecausing convection currents, heating the hot side to a uni-form high temperature (Fig. 2). The near neutrally buoyantdivider plate is moved axially by small actuators, with neg-ligible power consumption, to maintain the hot and coldsalt volumes required for continuous operation. As aresult, high-temperature salt can be provided even as theaverage temperature in the tank decreases (Fig. 3).

As the divider plate is lowered when the aperture is openand the tank is being heated by sunlight, colder salt frombelow moves past the annular clearance space between

1522 A.H. Slocum et al. / Solarthe barrier plate and tank wall to be reheated. Relativeblow-by salt velocities are slow (much less than 1 cm/sent temperature of 300 K, the exergetic mixing eciency(1 = no loss) is found as 0.935; this value approaches 1for higher temperature CSPonD systems. Mixing lossesare incorporated in the worst case estimates of receiver e-ciency listed as 0.6 in Table 2.

The salt-lled annular clearance between the loose t-ting divider plate and tank walls acts as a buer by limitingthermal gradients as the divider plate moves. Degradationof the tank walls due to thermal shock is reduced comparedto well-sealed barrier thermoclines with sharp thermalinterfaces (Copeland et al., 1984). The tank is internally

Fibrous insulation

Molten salt

Flexible alloy liner

Steel tank shell

Exterior Sheathing

Insulating refractory

brickgradients and shock. Cross-section adapted from Kolb (1993) andGabbrielli and Zamparelli (2009).

insulated with mortarless insulating refractory brick, whosethermal resistance and mass limits temperature swings atthe mild steel tank shell, as depicted in Fig. 4 (Kolb,1993). If needed, a thin corrugated alloy liner can beemployed which minimizes thermal shock and reduces ero-sion and spalling of the refractory brick in the tank; thisinternally-insulated design has been shown to be morecost-eective than a stainless steel tank with external insu-lation (Gabbrielli and Zamparelli, 2009). Metallurgical saltbaths have proven commercial super-duty alumina/silicaor pure silica reclay bricks adequately resist corrosionfrom molten nitrate or chloride mixtures (ASM, 1991).Recent tests have veried the cyclic stability of silica parti-cles in molten sodium and potassium nitrate at the temper-atures considered (Brosseau et al., 2005).

Two 90 arcs of heliostats with radius 390 m feeding twoapertures on a volumetric molten salt receiver 25 m diame-ter and 5 m deep containing 4500 tons (2500 m3) of 550 Cmolten nitrate salt could power a 4 MWe steam turbinecontinuously, 24/7 (7 h sunshine, 17 h storage) on a daily

A.H. Slocum et al. / Solar Enebasis. In addition, the heliostat eld is sized so that forevery 10 days of full sun, the system could also run foran addition 24 h on a cloudy day (Table 1). Conservatively,one can assume it takes about 16 m3/MWe/h of nitrate saltfor non-sunshine operation. The CSPonD system is ratedby continuous power production, not peak power as is typ-ical of traditional CSP systems without overnight storage.

Greater power cycle eciencies can be realized with theuse of larger steam turbines. This can be accomplished withlarger CSPonD receivers or several smaller units feeding acommon power block. However, many locales will favorthe use of CSP-generated power for peaking purposes, withgreatly reduced storage needs. As such, the same 2500 m3

storage volume can supply a 50 MWe turbine for 3.25 h

Table 1Design parameters for 4 MWe 24/7 nitrate salt CSPonD system with eithertwo 90 arc segment heliostat elds and two apertures or a single 135 arcsegment heliostat eld and a single aperture.

Site parameters

Solar altitude 80Distance receiver from base of hill 200 mHill angle 30Average daily insolation (24/7/365) 200 W/m2

Net average 24/7 power generation 4 MWe

CSPonD system parameters Number of heliostatarc segments

1 2Heliostat arc angle () 135 90Heliostat eld radius (m) 500 390Number of heliostat rows 88 74Total heliostat mirror area (m2) 139,357 141,312Unit aperture height (m) 10.3 8.9Unit aperture width (m) 12.9 11.1Total aperture area (m2) 133 199Salt tank diameter (m) 24.9 25.0

Salt tank depth (m) 5.0 5.0Salt required for 24/7 operation (metric tons) 4382 4512without additional solar input. After peak demand sub-sides, heat extraction can be stopped and nighttime lossesfrom the well-insulated receiver are minimal.

The receivers capture eciency can be dened as thefraction of incoming energy retained by the receiver usedto heat both the pond and the lid. Aperture size is driven bysystem sizing and input ux concentration and is used tocalculate geometrical view factors and radiative losses tothe environment. Other losses include conduction thoughthe tank walls and convection to the outside environment.In general, capture eciency increases with input ux con-centration as the system geometry approaches that of ablackbody. The self-healing nature of the molten salt sur-face tolerates much higher uxes than conventional tubebased receivers and can achieve higher eciencies asheliostat eld technology and achievable concentrationimproves.

High lid temperature leads to reduced capture eciencydue to large radiative and convective losses out of the aper-ture. However, a higher lid temperature reduces radiativeexchange from the salt pond to the lid eectively keepingmore energy in the salt. For example, a 240 C lid and550 C salt temps (e 0.9), roughly 18 kW/m2 is lost fromthe salt surface to the lid, 14% of the incoming power whenthe aforementioned receiver is on sun at the design point.With the aperture closed for 16 h, salt-lid radiative transferwould result in nearly 140 MW h of energy transferred tothe roof nearly one-quarter of the 600 MW h (equivalentto 180 MW h electric at gth = 0.30) storage capacity. Thisupper bound assumes a fully-exposed salt surface and aconstant salt surface temperature. However, high solar con-centration enables reductions in exposed salt surface area,further reducing losses and heat gain from the salt to thelid. Also, the pond free-surface will cool due to radiative,as well as internal convective heat transfer while o-sun.As a result, permanent heat transfer from the pond to thelid will decrease. Additionally, if energy was not extractedfrom the lid at a sucient rate, the lid temperature willincrease. Low lid thermal mass will result in quicker heatup and reduced nighttime losses. One possible operationalstrategy would be to slow or stop lid heat rejection towardsthe end of the day while still on sun, allowing the lid temper-ature to equilibrate with the salt, reducing radiative trans-fer. Permanent heat transfer losses are incorporated intothe receiver eciency estimates (0.60.9) in Table 2 forthe continuous operation CSPonD system examined (i.e.,both salt and lid heat rejection running 24/7).

Similarly, large beam down angles (i.e., steeper hillsides)are eective at directing more of the incoming energy intothe salt. A steep beam down angle has two eects: the illu-minated or projected area of the aperture on the horizontalmolten salt pool is decreased, reducing exposed salt areaand subsequent losses; and Fresnel reections o of the saltsurface to the lid are reduced. To avoid hot spots and min-imize system complexity, natural convection within the salt

rgy 85 (2011) 15191529 1523is needed. This can be designed into the pond using numer-ical methods (Nave, 2004). Preliminary results using the

of

its

/m2

e

t/M

t/M

t/M

e/M

ch a2

M 4.8 1.9M 0.2 0.1M 4.8 2.4M 8.6 3.5M/year 2.0 0.5

e/MWt 0.08 0.2249.4 18.5197,530 61,730

M 42.5 9.3ear 6132 7884h/year 24.5 31.5

M 61 17W h 0.33 0.07

Enefull variable-property NavierStokes equations to producedirect numerical simulations show thermal gradients can beused to create plumes to mix the salt.

Table 2Upper and lower bounds for predicted CSPonD levelized cost

Property Un

Rating and eciencies (input)

Average specic direct normal radiation kWNameplate capacity MWHeliostat overall eciency MWPond eciency MWHX eciency MWPower cycle eciency MWAvailability Capacity factor (1-Derate)Capital cost (input). Excluding contingencies, tax and land whi

Heliostat specic cost $/mPond cost $MHeat exchanger cost $MPower block cost $MUtilities, piping, site work $MOperating costs $M

Calculated properties

Overall eciency MWDirect normal radiation required MWHeliostat area required m2

Heliostat cost $MOperation h/yYearly electricity produced GWCapital cost $MSAM results $/k

1524 A.H. Slocum et al. / SolarThe CSPonD system could likely support higher powercycle temperatures should the need arise in the future.Future generation NaClKCl CSPonD designs could beconsidered for high temperature operation and the twozone tank could still be used. Most of the salt handlingand power-cycle technology required for a hot chloride saltoption for CSPonD has been partly developed for the mol-ten salt nuclear reactor program (Forsberg et al., 2007).Liquid salt pumps have been tested up to 6000 L/min attemperatures exceeding 700 C (Grindell et al., 1960;Williams, 2006). Multiple CSPonDs could feed a commoncentral power plant (Forsberg and Moses, 2009).

With regard to the stability of the salts, there is a largeindustrial experience base with molten salts used in the heattreatment of metals (Mehrkam, 1967; ASM, 1991). Thesehigh temperature salt baths are open to the atmosphereas metal components are moved in and out of the salt bath(Fig. 5). The heat treating industry has developed standardmethods to test the salt, using additives to control saltchemistry, and replacing the salt if the impurity levels aretoo high. The rate of impurity buildup will be much lowerfor CSPonD than for a heat treating bath with its dailythroughput of steel parts.

The use of quartz windows was considered to helpreduce radiation losses and eliminate mass transfer acrossthe aperture. However, as mentioned in literature and asobserved in our testing, quartz crucibles in contact withmany salts tend to lose their optical clarity with time (Lienergy.

Conservative Optimistic

0.25 0.304.0 4.0

Wt 0.5 0.6Wt 0.6 0.9Wt 0.9 1.0Wt 0.3 0.4

0.7 0.9

re included in SAM

215 150

rgy 85 (2011) 15191529and Dasgupta, 2000). The salts chosen as appropriate forCSPonD systems have limited toxicological and environ-mental eects should some condense outside the receiver;however, any system which regularly loses heat transfer u-ids (other than steam) to the environment will likely bedenied a permit. Fortunately, the salt bath heat treatingindustry provides guidance. Salt bath furnaces employ ven-tilation hoods, either oset from the salt surface or directlyoverhead to capture salt fumes at their source. The saltvapors are drawn into the intake, and condense on theintake plenum. The accumulated salt is scraped o duringroutine maintenance. A similar system can be installedabove the aperture of the CSPonD receiver, perhaps utiliz-ing an intake plenum with an automated salt scraping and

Fig. 5. Open air NaClKCl salt bath at 900 C for metal heat treating.(Picture taken at Metallurgical Solutions, Inc. in RI).

collection system. Salt vapor and thermal losses can also bemitigated with the use of air curtains installed across theaperture (Paxson, 2009).

Di Da= sin/ 2where Da is the aperture diameter and / is the nominalbeam-down angle measured from horizontal. A portionof the incoming concentrated light is reected o the mol-ten saltair interface. This reected fraction can be calcu-lated from the Fresnel equations describing light as itmoves between media of dierent refractive indices. Thereection coecients for s- and p-polarized light are

Rs sinht hisinht hi

23

Rp tanht hitanht hi

24

The angles that the incident, reected and refracted raysmake to the normal of the interface are given as hi, hr

A.H. Slocum et al. / Solar Ene2.1. Volumetric absorption in molten salts

The light attenuation characteristics of the molten saltare critical as they strongly aect temperature gradientswithin the salt and overall system design. For example, ifthe salt is too attenuating, in large deep ponds createdfor a large amount of storage, forced circulation of the saltcould be required to prevent overheating of the top surfaceof the salt. Conversely, if the salt is found to be too trans-parent, overheating of the divider plate could occur if it isbrought too close to the surface. The attenuation of the saltcould be increased with the addition of nanoparticles to thesalt, or conversely, impurities that build up with time couldaect the behavior of the salt. The attenuation characteris-tics of the CSPonD candidate molten salts are currently notavailable. Accordingly, the eorts undertaken to obtainthis data are detailed here.

2.1.1. Attenuation measurements in molten salt

Light attenuation in a semi-transparent medium isdescribed by the attenuation (or extinction) coecient, a(cm1), which represents the probability per unit lengththat a photon will be removed from the incident beam,either by absorption or scattering (Passerini, 2010). Theattenuation coecient depends on the photon energy, orwavelength, and also on the temperature of the medium.The intensity transmitted through a layer of material ofthickness d is related to the attenuation coecient by thefollowing equation:

I I0ead 1where I and I0 are the transmitted and incident intensities,respectively. Eq. (1) can be used to calculate the attenua-tion coecient by measuring the intensity of light transmit-ted through known depths of a molten salt.

A furnace-based apparatus was designed and built tomeasure optical properties of a number of candidatemolten salts at various temperatures and wavelengths(Passerini, 2010). It is capable of operating from roomtemperature up to 1100 C. The apparatus and methodwere validated against published data by comparing resultsobtained for deionized water at room temperature andHitec salt1 from 250 to 500 C (Drotning, 1978; Smithand Baker, 1981). Due to the light signal intensity distribu-tion, reliable data were obtained between 400 nm and800 nm, while for shorter and larger wavelengths the signalto noise ratio drops signicantly, making quantitativeevaluation of the attenuation coecients impossible.1 Costal Hitec is a name brand, ACS reagent grade chemicals were usedat the same composition as the Hitec product.For illustrative purposes, the measured attenuationcoecient of a 60/40 wt.% sodiumpotassium nitrate mix-ture at 350 C has been applied to the reference solar irra-diance plot of ASTM G173-03. Fig. 6 shows the result ofapplying the measured attenuation to this reference solarspectrum for dierent molten salt depths. At a depth oftwo meters, 93% of the total solar energy is absorbed bythe candidate nitrate salt in the 400800 nm range. It isimportant to note that 55% of the terrestrial solar insola-tion lies in the 400800 nm range. Extrapolation of thisdata beyond the 400800 nm range is dicult. However,it is expected that highly absorbing regions would be inthe infrared region at wavelengths greater than 2.5 lm,where the solar irradiance expires. Accordingly, some rstorder extrapolation could be performed.

2.1.2. Irradiance distribution in receiver

For a central heliostat located on the optical axis of thereceiver aperture, the input ux will be restricted to anellipse of major axis Di projected on the salt surface, calcu-lated as

0

0.2

0.4

0.6

0.8

1

300 600 900 1200 1500 1800 2100 2400

Nor

mal

ized

Irra

dian

ce

Wavelength (nm)

Reference: AM1.5

0.25 m Salt Depth

0.75 m Salt Depth

1 m Salt Depth

2 m Salt Depth

3 m Salt Depth

Fig. 6. Solar irradiance attenuation for dierent molten salt depths. Thisgure was produced using experimentally measured attenuation coe-cients for a 60/40 wt.% sodiumpotassium nitrate mixture at 350 C in thewavelength range of 400800 nm, the range where experimental measure-ments are possible.

rgy 85 (2011) 15191529 1525and ht, respectively, and related by Snells Law and thelaw of reection.

nsalt sin ht nair sin hi 5hr hi 6where

hi 90 / 7Since the incident light is unpolarized, containing an equalmix of s- and p-polarizations, the reection coecient, R, is

R Rs Rp=2 8Eq. (8) can be used to nd the percentage of the incidentenergy which is reected onto the lid. For the sample caseof nitrate solar salt with / = 20 and nsalt = 1.413 (Bloomand Rhodes, 1956), 84.7% of the incoming radiation re-fracts into the salt while 15.3% reects onto the lid. Theintensity of the refracted light at various salt depths is thenfound with Eq. (1). Fig. 7 illustrates the incoming solarirradiance distribution for a centralized heliostat aimed at

on a small scale receiver tank equipped with a movable

the hottest region of the receiver is the top surface of divi-der plate. This creates natural convection cells in the topregion thereby preventing overheating of the top surfaceand maximizing thermal storage in a given volume of salt.Fig. 9 depicts the convection currents within an alternatetest receiver detected with a lab-built particle image veloc-imetry (PIV) system. Indeed, the light penetrating the saltcauses heating deep within the sample and convective mix-ing readily occurs to prevent hotspots.

Due to packaging constraints, the small-scale dividerplate test bed utilized a 1=4

00 (6.35 mm) rigid silica insulatingboard axed to the underside of the stainless steel dividerplate. Full scale systems would utilize signicantly moreinsulation, on the order of 1020 cm, providing a greaterthermal resistance and limiting heat transfer while provid-ing the design value of 300 K across the divider plate. In alarger and deeper system, the upper surface of the dividerplate will receive less radiation than this small scale exper-iment. However, with full-scale insulation on the dividerplate, conduction through the bottom plate will decrease.It is expected that natural convection into the upper hotsalt region will remain of the same magnitude, and a tem-perature inversion should occur in the hot salt region

Temperature (C)

Fig. 8. Temperature distribution of CSPonD receiver optically heated bythe 10.5 kW MIT CSP solar simulator; inset depicts stainless steel (type316 L, 28 cm inner diameter 8 cm high, 7 L solar salt capacity) testreceiver. The movable divider plate (3.2 mm thick stainless steel disk witha 6.4 mm thick layer of rigid silica insulation board axed to theunderside) and is denoted by the shaded gray box located near the saltsurface (A) and tank bottom (B) of the test receiver.

1526 A.H. Slocum et al. / Solar Enedivider plate, designed to partition the volumetric moltensalt receiver into two thermally separated regions (Fig. 8).

The divider plate provides excellent thermal separationbetween the hot upper and cold bottom sections. The barestainless steel top surface of the divider plate absorbs muchmore energy than the relatively transparent salt; as a result,

Fig. 7. Representative solar irradiance distribution for a single heliostatthe receiver, simulated using MATLAB. An entire heliostateld array can be aimed to create a circular illuminationdistribution on the salt surface; much of the incoming radi-ation will be absorbed near the top of the salt surface andtowards the rear of the tank. Hence, it is logical to extracthigh temperature salt in this location for the power cycle.

2.1.3. Solar simulator testing of a volumetric salt receiver

To test the CSPonD concept, a high-ux large-area solarsimulator was designed to achieve output uxes greaterthan 60 kW/m2 (Codd et al., 2010). Tests were conductedaimed at a CSPonD receiver. A complete heliostat eld can be aimed tocreate a circular illumination distribution on the salt surface.0

1

2

3

4

5

6

7

8

280 300 320 340 360 380

Hei

ght (

cm)

Temperature ( C)

t = 2000 s 4000 60008000

divider plate

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

280 300 320 340 360 380H

eigh

t (cm

)

divider plate

t = 2000 s 40006000

8000

(A)

(B)

rgy 85 (2011) 15191529accordingly the Nusselt number will remain similar to thisexperiment. Regrettably, the 60-sun solar simulator was

across the divider plate.

aretwoovem

EneHowever, these small scale experiments cannot provideinsight to how a full scale machine would truly work.Hence, numerical modeling is required to design the fullscale tank and divider plate system, and then testing of adeep tank system.

2.2. Site selection and heliostat placement

There is precedence for the location of heliostats on ahillside to direct sunlight to a secondary reector, thenredirecting the power to a receiver on the ground (Trombeand Le PhatVinh, 1973); however, up to 10% of the energynot capable of heating the salt above 380 C. This was to beexpected with high conductive, surface radiative and con-vective losses for this low-ux test receiver. Full scale sys-tems will utilize much greater ux levels and havesignicantly reduced system losses, which will increasethe hot salt temperatures and the temperature dierence

Fig. 9. Molten salt convection cell monitoring using PIV. Notice that thereuid movement within the receiver. The inset on the right is composed ofhave shifted from frame-to-frame; these can be used to characterize uid mreader is referred to the web version of this article.)

A.H. Slocum et al. / Solaris lost with each additional reection, not to mention high-ux secondary mirror cooling concerns, operation andinstallation costs. Meanwhile, there appears to have beena land rush for acquiring rights to at, sunny land per-ceived to be needed for other types of solar power systems,which have increased the overall costs of traditional CSPsystems. The system presented here thus reects the solarenergy from a heliostat eld on a hillside directly into areceiver. In the northern hemisphere, a south-facing hillsideeld allows for direct beam-down entry into the molten saltpond as shown in Fig. 1A. These congurations allow forCSP collector elds to be built on otherwise undevelopable,steep terrain hence reducing system costs. Methods used byutility companies for emplacing utility poles on moderatelysteep terrain can be used for heliostat installation, andautomated spray systems can be utilized for cleaning themirrors. Indeed, one of our important conclusions is con-ventional power tower systems could benet from hillsidemounted heliostats. A numerical tool was developed toevaluate potential sites for beam up and beam downCSPonD congurations (Noone et al., 2011). For two casestudies, White Sands, NM and China Lake, CA, the areainvestigated is roughly 10,000 km2 and optimal receiverlocations for the beam-down conguration have ecienciesof 70% and 68%, respectively. In the beam-up congura-tion, both case studies have optimal receiver locations witheciencies of 77%. These eciencies take into accountcosine eciency, shading and blocking losses due to theterrain; optical losses of the heliostats are not included inthese results. However, heliostat overall eciency, givenin Table 2 as 0.50.6, includes heliostat optical lossesincluding reectivity, mirror shape and tracking errors.

For each case study, one result is included to illustratean optimal location for a pond receiver for each of two sce-narios, (i) reecting the sunlight up the hill into a secondaryreector at the receiver with an optical eciency of 0.9, and(ii) reecting the sunlight down the hill so as to not requirea second reection. Similar results were obtained for bothsites, and a summary of the White Sands results are pre-

particles in the molten salt that reect light which can be used to determineconsecutive time frames (blue and green). The arrows show particles thatent. (For interpretation of the references to color in this gure legend, thergy 85 (2011) 15191529 1527sented in Fig. 10.Sites with highest eciency from the White Sands area

were shown in Fig. 1. In these sites, the heliostat elds arenorth (right) of the receiver, represented in the image forvisualization of the eld extents only, not an actual heliostatlayout. The eld eciency of the site with the secondaryreector is calculated to be 77% and the site without is70%, a dierence of 7% despite the 10% loss associated withthe grazing angle reection o the inner surface of the lid.Assuming that 15% of the land can be utilized, and of thisland 30% is covered by heliostats, a solar-to-electric e-ciency of 22%, and a 24/7 average solar insolation of200 W/m2, the White Sands site could provide 20 GWe ofpower 24/7. Similar results are obtained for China Lake.

2.3. Economics

The uncertainty associated with new technologies is verylarge and therefore a conservative as well as an optimisticeconomic analysis was conducted. The Solar Advisory

er snden eanaated

EneModel (SAM) by NREL was used along with the economicassumptions by US Department of Energy for CallDE-FOA-0000104 set for 2020, including some incentives(Sargent and Lundy, 2003).

The optimistic scenario gives an economical levelizedcost of 0.07 $/kW h, while a conservative estimate is cur-rently prohibitively high with a value of 0.33 $/kW h(Table 2). It should however be noted that this gure ishigh compared to present fossil fuel costs, not necessarilyhigh compared to conventional CSP systems. The rangeof levelized energy cost clearly demonstrates the need forfurther research, in particular with regards to (i) thedetailed design of the pond and lid along with selectionof salt and materials, (ii) the best use of the land availablevia optimal placement of heliostats and (iii) the optimalintegration of the pond and lid with the power cycle (andpotentially a cogeneration scheme).Fig. 10. Density overlay indicating areas of high eciency CSPonD receivlocations corresponding to high annual average eld eciencies are in red aRadar Topography Mission (SRTM), with one arc-second resolution betwetopocentric plane (Farr et al., 2007). Due to the coarseness of the data, theapproximating eld eciency. The instantaneous beam irradiation is calculdirect beam irradiance under cloudless skies (ERSI, 2008; Google, 2010).

1528 A.H. Slocum et al. / Solar3. Conclusions and future work

A CSP system with integral storage has been presentedwhere hillside mounted heliostats direct sunlight into a vol-umetric absorption molten salt receiver either near the baseof the hill where the sunlight can directly penetrate the salt,or at the top of the hill where the light is redirected o thelid of the receiver before it enters the salt. In both cases, saltvapor condenses on the inside surface of the lid to form aself-healing reective surface.

The concentrated light penetrates and is absorbed in thereceiver by molten salt through a distance of 45 m whichallows for high solar uxes and is structurally insensitive tocloud cover transients that aect other CSP systems. Thereceiver has a relatively small aperture with lower second-ary heat losses and avoids thermal fatigue associated withboiler tube-type receivers while achieving high tempera-tures needed for ecient power generation. In addition,the receiver volume also acts as the thermal storage vol-ume. Hot salt is pumped from the top of the tank througha heat exchanger and then back into the bottom of thetank. An insulated plate provides an additional thermalbarrier between the thermally stratied hot and cold layerswithin the tank, and the barrier is moved axially up anddown to provide high temperature thermal energy even asthe average temperature of the salt in the tank decreaseswhen the sun is not shining.

Fundamental measurements of the optical absorbanceproperties of molten salt sand bench level experiments indi-cate viability of the concept. Analysis of hillsides at twosouthwestern government land sites shows good potentialfor CSP system development. An economic analysis usingNRELs Solar Advisor program indicates a levelized costof electricity of $0.070.33/kW h.

The next step in the research is to design a 20100 kWttest receiver that has an aperture size to receive light from atypical concentrating heliostat, so a commercial array canbe easily modied and used without needing additional

ites in White Sands, NM: (A) beam down and (B) beam up sites. Receiverregions of poor eciency in blue. Digital elevation data is from the Shuttlelevation data points (SRTM1 v2.1), corresponding to roughly 30 m on thelysis is regarded as a simplied model used for locating potential sites andusing version one of the Meteorological Radiation Model (MRM v1) forrgy 85 (2011) 15191529concentrating optics at the aperture. This receiver would,however, be designed with the full anticipated depth of alarger system, so the optical penetration and convectivemixing properties anticipated for the MWe sized CSPonDsystem can be evaluated.

Acknowledgments

This work is part of an interdisciplinary collaborationbetween the Cyprus Institute, the University of Illinois atUrbana Champaign, the Electricity Authority of Cyprus,and the Massachusetts Institute of Technology. Generousgraduate student fellowships were provided by theChesonis Family Foundation and the Bill and MelindaGates Foundation. The authors would also like to thankProf. Jerey M. Gordon of Ben-Gurion University of theNegev and Dr. Steve Fantone of Optikos, Inc. for theiroptical engineering insights, and Prof. Alan Hatton ofMIT for his evaluation of the viability of a nanoparticle-based approach to light absorption in molten salts.

A.H. Slocum et al. / Solar EneReferences

ASM, 1991. ASM Handbook. Heat Treating, 10th ed. vol. 4. ASMInternational.

Bassett, I., Derrick, G., 1981. An upper bound on the eciency of a cavityabsorber. Opt. Acta (UK) 28, 5763.

Bloom, H., Rhodes, D.C., 1956. Salt mixtures. Part 2. The refractive indexof molten nitrate mixtures and their molar refractivities. J. Phys.Chem. 60, 791793.

Bohn, M.S., 1987. Experimental investigation of the direct absorptionreceiver concept. Energy 12, 227233.

Brosseau, D., Kelton, J., Ray, D., Edgar, M., Chisman, K., Emms, B.,2005. Testing of thermocline ller materials and molten-salt heattransfer uids for thermal energy storage systems in parabolic troughpower plants. Trans. ASME, J. Sol. Energy Eng. (USA) 127, 109116.

Codd, D.S., Carlson, A., Rees, J., Slocum, A.H., 2010. A low cost highux solar simulator. Sol. Energy 84, 22022212.

Copeland, R., Green, J., 1983. Raft thermocline thermal storage. In:Proceedings of the 18th Intersociety Energy Conversion EngineeringConference, 2126 August 1983. AIChE, New York, NY, USA, pp.18011805.

Copeland, R., West, R., Kreith, F., 1984. Thermal energy storage at900 C. In: 19th Intersociety Energy Conversion Engineering Confer-ence (Cat. No. 84CH2101-4), 1924 August, 1984. American Nucl.Soc. LaGrange Park, IL, USA, pp. 11711175.

Derrick, G., Bassett, I., 1985. Optimization of imperfect diuse reectors.Opt. Acta (UK) 32, 313328.

Drotning, W., 1978. Optical properties of solar-absorbing oxide particlessuspended in a molten salt heat transfer uid. Sol. Energy (USA) 20,313319.

Epstein, M., Segal, A., Yogev, A., 1999. A molten salt system with aground base-integrated solar receiver storage tank. Le Journal dePhysique IV, 9.

ERSI, 2008. ESRI Data and Maps, ArcMap v9.3 (GIS software).European Commision, 2005. EUR 21615 SOLGATE Solar Hybrid Gas

Turbine Electric Power System. Oce for Ocial Publications of theEuropean Communities, Luxembourg.

Farr, T.G., Rosen, P.A., Caro, E., Crippen, R., Duren, R., Hensley, S.,Kobrick, M., Paller, M., Rodriguez, E., Roth, L., Seal, D., Shaer, S.,Shimada, J., Umland, J., Werner, M., Oskin, M., Burbank, D.,Alsdorf, D., 2007. The Shuttle Radar Topography Mission [WWWDocument]. Reviews of Geophysics. .

Forsberg, C.W., Moses, D.L., 2009. Safeguards Challenges for Pebble-BedReactors (PBRs): Peoples Republic of China (PRC) (No. ORNL/TM-2008/229). Oak Ridge National Laboratory (ORNL).

Forsberg, C.W., Renault, C., Le, B., Merle-Lucotte, E., Ignatiev, V., 2007.Les applications des selsliquideset les reacteurs a` selsfondus = Liquidsalt applications and molten salt reactors. Revue generalenucleaire,6371.

Gabbrielli, R., Zamparelli, C., 2009. Optimal design of a molten saltthermal storage tank for parabolic trough solar power plants. J. Sol.Energy Eng. 131, 041001.

Ghobeity, A., Mitsos, A., 2009. Optimal use of solar thermal energy forcombined power generation and water desalination. In: Proceedings ofDistributed Renewable Energy Sources in the Mediterranean Region(DISTRES 2009). Presented at the DISTRES 2009, Nicosia, CY.

Ghobeity, A., Noone, C.J., Papanicolas, C.N., Mitsos, A., submitted forpublication. Optimal Time-Invariant Operation of a Power and WaterCogeneration Solar-Thermal Plant. Solar Energy.

Google, 2010. Google Earth [WWW Document]. .

Goosen, M.F., Sablani, S.S., Al-Maskari, S.S., Al-Belushi, R.H., Wilf, M.,2002. Eect of feed temperature on permeate ux and mass transfercoecient in spiral-wound reverse osmosis systems1. Desalination144, 367372.Grindell, A.G., Boudreau, W.F., Savage, H.W., 1960. Development ofcentrifugal pumps for operation with liquid metals and molten salts at11001500 F. Nucl. Sci. Eng. 7.

Herrmann, U., Kelly, B., Price, H., 2004. Two-tank molten salt storage forparabolic trough solar power plants. Energy 29, 883893.

Kolb, G.J., 1993. Thermal cycling of thermal energy storage tanksproposed for the solar two central receiver power plant. In: ASMEInternational Solar Energy Conference, April 4, 1993April 9, 1993,Solar Engineering. ASME, Washington, DC, USA, pp. 397405.

Lata, J.M., Rodrguez, M., de Lara, M.A., 2008. High ux centralreceivers of molten salts for the new generation of commercial stand-alone solar power plants. J. Sol. Energy Eng. 130, 021002.

LeBarre, S., 2010. The Future of Green Architecture: A Live-In PowerPlant. Popular Science (July), 6263.

Li, J., Dasgupta, P.K., 2000. A simple instrument for ultraviolet-visibleabsorption spectrophotometry in high temperature molten salt media.Rev. Sci. Instrum. 71, 2283.

Mehrkam, Q.D., 1967. An Introduction to Salt Bath Heat Treating.Tooling & Production June/July 1967.

Nave, J.C., 2004. Direct Numerical Simulation of Liquid Films. Ph.D.Thesis, UCSB.

Noone, C.J., Ghobeity, A., Slocum, A.H., Tzamtzis, G., Mitsos, A., 2011.Site selection for hillside central receiver solar thermal plants. SolarEnergy. 85, 839848.

Pacheco, J.E., Showalter, S.K., Kolb, W.J., 2002. Development of amolten-salt thermocline thermal storage system for parabolic troughplants. J. Sol. Energy Eng. 124, 153.

Passerini, S., 2010. Optical and Chemical Properties of Molten SaltMixtures for Use in High Temperature Power Systems. SM Thesis,MIT.

Paxson, A.T., 2009. Design and Validation of an Air Window for aMolten Salt Solar Thermal Receiver. SB Thesis, MIT.

Rabl, A., 1976. Tower reector for solar power plant. Sol. Energy 18, 269271.

Reilly, H.E., Kolb, G.J., 2001. An Evaluation of Molten-Salt PowerTowers Including Results of the Solar Two Project. Sandia NationalLabs, Albuquerque, NM, Livermore, CA.

Sargent, Lundy, L.L.C., 2003. Assessment of Parabolic Trough and PowerTower Solar Technology Cost and Performance Forecasts (No. SR-550-34440). NREL.

Segal, A., Epstein, M., 2003. Solar ground reformer. Sol. Energy 75, 479490.

Semenyuk, L.G., 1983. Exergy loss on mixing working bodies withdierent temperatures. J. Eng. Phys. Thermophys. 44, 617619.

Skinrood, A.C., Brumleve, T.D., Schafer, C.T., Yokomizo, C.T., LeonardJr., C.M., 1974. Status report on a high temperature solar energy system(No. SAND74-8017). Sandia Labs., Albuquerque, N. Mex. (USA).

Slocum, A.H., Buongiorno, J., Forsberg, C.W., Codd, D.S., Paxson, A.T.,2010. Concentrated Solar Power System. PCT Patent ApplicationPCT/US10/49474.

Smith, R.C., Baker, K.S., 1981. Optical properties of the clearest naturalwaters (200800 nm). Appl. opt. 20, 177184.

Trombe, F., Le PhatVinh, A., 1973. Thousand kW solar furnace, built bythe national center of scientic research, in Odeillo (France). Sol.Energy 15, 5761.

Viebahn, P., Kronshage, S., Lechon, Y., 2008. Deliverable n 12.2-RS Ia.Final report on technical data, costs, and life cycle inventories of solarthermal power plants (Project No: 502687). NEEDS New EnergyExternalities Developments for Sustainability.

Williams, D.F., 2006. Assessment of Candidate Molten Salt Coolants forthe NGNP/NHI Heat-Transfer Loop. ORNL/TM-2006/69, OakRidge National Laboratory, Oak Ridge, Tennessee.

Yogev, A., 1997. Heat storage device. US Patent 5685289.Yogev, A., Kribus, A., Epstein, M., Kogan, A., 1998. Solar tower reector

systems: a new approach for high-temperature solar plants. Int. J.Hydrogen Energy 23, 239245.

rgy 85 (2011) 15191529 1529

Concentrated solar power on demand1 Background2 CSPonD: collocated receiver and storage system2.1 Volumetric absorption in molten salts2.1.1 Attenuation measurements in molten salt2.1.2 Irradiance distribution in receiver2.1.3 Solar simulator testing of a volumetric salt receiver

2.2 Site selection and heliostat placement2.3 Economics

3 Conclusions and future workAcknowledgmentsReferences