dynamic simulation of concentrating solar power plant and two-tanks direct thermal energy storage

at SciVerse ScienceDirect

Energy xxx (2013) 1e9

Contents lists available

Energy

journal homepage: www.elsevier .com/locate/energy

Dynamic simulation of concentrating solar power plant and two-tanks directthermal energy storageq

Flavio Manenti*, Zohreh Ravaghi-ArdebiliPolitecnico di Milano, Dipartimento di Chimica, Materiali e Ingegneria Chimica “Giulio Natta”, Piazza Leonardo da Vinci 32, 20133 Milano, Italy

a r t i c l e i n f o

Article history:Received 29 October 2012Received in revised form31 January 2013Accepted 1 February 2013Available online xxx

Keywords:Archimede plantConcentrating solarEconomizer modelDynamic simulationThermal energy storageDirect storage

q Prepared for Energy on invitation.* Corresponding author. Tel.: þ39 (0)2 2399 3273;

E-mail address: [email protected] (F. Mane

0360-5442/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.energy.2013.02.001

Please cite this article in press as: Manenti Fthermal energy storage, Energy (2013), http

a b s t r a c t

The discontinuous nature of the solar energy forces to study the dynamic behavior of solar plants tocharacterize their operations, to deepen their process understanding and to improve the performanceand maintenance. The present paper focuses on the dynamic simulation and control of concentratingsolar plants with the aim to define a reasonably simplified layout as well as to highlight the main issuesto characterize the process dynamics of these energy systems and their related energy storages. Detailedfirst-principles mathematical models of key unit operations are developed, implemented, and integratedinto commercial codes to improve the reliability of the plant dynamic simulation as well as the previsionaccuracy. The case of Archimede concentrating solar power plant with the related two-tanks directthermal energy storage technology is investigated.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Solar energy for land applications, but in practice severalrenewable sources, is a discontinuous source; on the other hand,although with significant daily, weekly and seasonal dynamics, theenergy thirsty of industrialized countries can be considered ofcontinuous nature. To meet source availability andmarket demand,there is the need to use energy storage options [1] and technologies[2], such as thermal energy storages [3] or thermochemical storages[4], or to switch to the partial or total production of chemicalcommodities such as biofuels [5] or hydrogen [6] to quote a coupleand performance evaluation [7] and techno-economical assess-ments are usually needed according to the location and re-quirements (see for instance [8]). Energy storage for solar plantsmeans to accumulate daytime a certain quantity of energy to beused during the night [9] or during unfavorable meteorologicalconditions [10].

Speaking about energy storages and discontinuous nature ofpower sources [11], the only way to study the solar energy pro-cesses as well as to improve their effectiveness and performancesfrom the process viewpoint is to pass through the dynamic

fax: þ39 (0)2 7063 8173.nti).

All rights reserved.

, Ravaghi-Ardebili Z, Dynamic://dx.doi.org/10.1016/j.energy

modeling and simulation especially for the systems involvingstrong dynamics like the concentrating solar plants. Several dy-namic simulations were performed to improve the knowledge andto well-assess the effectiveness of solar plants facing design [12],numerical simulation [13], and economic issues [14] and tomonitor[15], predict [16] and control [17] the thermal energy storages.Unfortunately, they are not based on well-established commercialdynamic simulators containing detailed models for such plants,since linear-parabolic solar troughs, molten salts mixtures, moltensalts steam generators, and tanks for energy storages are signifi-cantly different from the conventional units adopted elsewhere andtraditionally included in the model library for process simulations(see for example ref. [18] for oil & gas processes).

Thus, there are two possibilities: develop by new the models ortransform the existing models and make them reasonably accuratefor our simulation. If the unavailability of dedicated models willunavoidably lead to their definition by new and, if useful, to theirimplementation in existing commercial packages in the next future,this first research activity wants to give an overview of the diffi-culties behind the dynamic simulation of concentrating solar plantsthrough the existing tools. Thus, the key-point is to get a verysimplified, but at the same time performing and reasonably accu-rate, dynamic simulation. Such a simplified simulation is also usefulto assess the feasibility and reliability of several very appealingsolutions that are going to be installed soon in the solar field such asthe real-time performance monitoring of solar panels and units or

simulation of concentrating solar power plant and two-tanks direct.2013.02.001

Delta:1_given name

Delta:1_surname

mailto:[email protected]

www.sciencedirect.com/science/journal/03605442

http://www.elsevier.com/locate/energy

http://dx.doi.org/10.1016/j.energy.2013.02.001



F. Manenti, Z. Ravaghi-Ardebili / Energy xxx (2013) 1e92

the optimization of plant transients (i.e. startup and shutdown) forunit safeguard and plant safety.

The selected case-study is the Archimede solar plant, which isoperating in the south of Italy (Siracusa, Sicily), and it is described inSection 2. The detailed first-principles model for the economizerunit is given in Section 3. The dynamic simulation and the controlscheme are jointly described in Section 4. Considerations on thethermal energy storage for concentrating solar plants and thegeneral discussion are given in Sections 5 and 6, respectively.

2. Archimede concentrating solar power plant description

The Archimede Concentrating Solar Power (ACSP) plant islocated in Sicily (Italy) and schematically represented in Fig. 1; itconsists of two tanks for molten salts storage, a series of linear-parabolic solar panels and a steam generator with the associatedheat exchange train. The general flowsheet representing the dy-namic simulation of the ACSP plant, which the process descriptionhereinafter is referred to, is reported in Fig. 2.

Starting from the tank SALT, the molten salt is fed to the plantduring the startup procedure. It is fed to the tank COLD_TANKthrough the electric motor M1, the pump P1, and the valve XV1 thatare positioned on the startup line. Such a line is operating onlyduring the plant startup since no make-up of molten salts is usuallyneeded in the short/medium terms. Once the appropriate quantityof molten salts is entered, the molten salts start flowing from thetank COLD_TANK to the tank HOT_TANK passing through the pumpP2, the valve XV2, which is regulated by the flow controller PID1,and the concentrating solar troughs, modeled by the pipe modelPIP1. The solar energy HS1.Q is provided as external duty. This is thetypical daytime operation and we will refer to this ACSP plantsection as the solar line in the following. During the night, the solarline is off, whereas the ACSP plant generation line, from HOT_TANKto COLD_TANK, is always operating at the conditions imposed bythe electric energy market demand for just-in-time renewablepower supply tomatch industry expectations as above. For the sakeof simplicity, a constant market demand is assumed in this work.

The dynamic simulation is theoretically able to optimize thepower generation and to improve the net margins of ACSP plant bylooking at the market demand in real-time and automaticallyselecting the best setpoints for the power generation with relevanteconomic benefits as demonstrated in other fields [19,20,21]. As thediscontinuous nature of ACSP offers good possibilities of improve-ment, this point will be part of the future developments.

Four heat exchangers are placed on the generation line.Following the flow of molten salts, we encounter the SECOND SU-PERHEATER, the FIRST SUPERHEATER, the BOILER for steam gen-eration, and the ECONOMIZER. The molten salts temperaturedecreases in the flow direction. Conversely, the water enters fromthe tank WATER, passes through the valve XV6, is preheated in the

Fig. 1. ACSP qualitatively layout.

Please cite this article in press as: Manenti F, Ravaghi-Ardebili Z, Dynamithermal energy storage, Energy (2013), http://dx.doi.org/10.1016/j.energy

economizer, changes the phase in the boiler and the steam is su-perheated in the first and second superheaters before entering theturbine for power generation. Turbine efficiency is assigned:h¼ 37.5%. Please note that to simulate the boiler with a tube bundlecompletely submerged in the boiling water, it is necessary to useseparate units with the heat stream that links them. The plant isdesigned to produce 4.7 MW. The vapor flow is estimated in21.63 kg/s. 135 kg/s of hot molten salts have to pass through theheat exchanger train, under the ideal assumption of its maximumperformance. The heat exchanger train has an exchange area of3831 m2 with the overall heat transfer coefficient equal to 0.58 kW/m2/K. About the heating operation of the molten salts, the receiversare 0.07 m diameter tubes and the total length installed is about5400 m (9 strings of 6 collectors each and the length of one col-lector is about 100 m). The direct solar radiation is approximately1.9 kW/m2 in Sicily. Two tanks for the thermal storage are: 13 mheight and 29 m diameter the cold tank; 10 m height and 24 mdiameter the hot tank (source: www.enel.it). It is known that thecondensed water after the turbine is recycled back as inlet water inthe steam/water loop. In the proposed simulation, this loop isremoved to improve stability, flexibility and controllability of theoverall dynamic simulation, but the consistency of the inlet andoutlet water temperatures is preserved (243 �C).

According to Stephanopoulos’s directives, the simplest controlscheme that does the work is the best control scheme [22]. In thiscase, two proportional-integral flow controllers, PID1 for the solarline and PID2 for the generation line, and a proportional-integraltemperature controller, PID3, for the steam generation are imple-mented to manage the overall system, beyond the relevance tomonitor the minimum temperature of molten salts for loss of assets[23].

3. Detailed models for the ACSP plant

There are several unit operations in the concentrating solarpower plants that do not find yet a valid and reliable correspon-dence in the commercial dynamic simulators of process engineer-ing, usually adopted to design, operate and optimize chemical andoil&gas processes. This is a hard limitation if we think that thechemical process integration and energy intensification usuallypass through the process simulations for design and verificationpurposes. This limitation appears harder if we think that theexisting solutions for the dynamic simulation of concentrating solarplants are difficult to integrate with existing and widespread so-lutions for chemical and process engineering, making infeasible orvery problematic the feasibility studies dealing with the renewableenergy intensification of traditional chemical processes.

Actually, solar troughs, molten salts storage tanks, and heatexchangers do not have dedicated units in the common commercialpackages and especially in the tools for dynamic simulations, whichare needed to study and assess the effectiveness of energy storages.For this reason, the existing packages for dynamic simulation canbe considered as important structures to support complex flow-sheeting and detailed process analysis, but they need certainimprovement inmodeling to handle certain peculiarities of the unitoperations typical of solar power plants. Some units have beenalready investigated in the literature, whereas some others are notand this paper is aimed at bridging the gap presenting the detaileddynamic modeling of the economizer of the generation line of theACSP plant.

The economizer is a key unit operation in every concentratingsolar plant since it refines the lowest temperature of molten saltswithin the overall plant. In practice, the temperature of molten saltsoutflowing the process side of the economizer is always measured,monitored and controlled in order to prevent too low temperatures

c simulation of concentrating solar power plant and two-tanks direct.2013.02.001

http://www.enel.it

Fig. 2. ACSP flowsheet for dynamic simulation.

F. Manenti, Z. Ravaghi-Ardebili / Energy xxx (2013) 1e9 3

of molten salts and, therefore, total or local crystallization of such aheat thermal fluid. The economizer is the last unit of the heat ex-change network in the molten salts side and it has water to bepreheated in the shell side. It means that no phase changes arepresent in the unit. It is reasonable to assume:

� negligible radial gradients in the tube bundle and single tube;� cold fluid (water) in the shell side;� heat transferred from molten salts (T) to the metal wall (TM)and, next, to the cooling water;

� negligible potential and kinetic energy;� simplified forms for internal energy and enthalpy;� negligible diffusive flow with respect to the bulk flow.

Apart from the distributed nature of the heat exchanger thatshall be discussed later, the energy balance depends on the inletand outlet convection terms and heat losses to the ambient:

dHdt

¼ Hin � Hout þ Qs (1)

d�m$bh�dt

¼ m$bhin �m$bhout þ Qs (2)

d�m$cp$T

�dt

¼ m$cp;in$�Tin � Trif

��m$cp;out$

�Tout � Trif

�þ Qs

(3)

Since mass and specific heat can be considered constant andassuming the system perfectly mixed, the following general form isobtained:

Please cite this article in press as: Manenti F, Ravaghi-Ardebili Z, Dynamicthermal energy storage, Energy (2013), http://dx.doi.org/10.1016/j.energy

m$cpdTdt

¼ m$cp;out$ðTin � ToutÞ þ Qs (4)

Conversely, a more detailed model must be used to match atbest the relevant task of economizer, which is the preservation ofthe outlet temperature of molten salts at 290 �C with high reli-ability. Thus, the incremental typemathematical model for thermo-hydraulic calculations of the shell and tube exchangers is selectedaccording to the HTRI (Xist) standard and developed. It means thatthe unit is discretized in a certain number of volumes and specificrelationships (based on flow heat exchange conditions) are appliedfor each of them. In the acquaintance of the author, the proposedHTRI standard-based dynamic model is implemented for the firsttime:

� The economizer has 9 baffles and it is split into 10 elementsalong the longitudinal axis (Fig. 3). Each element correspondsto the volume between two diaphragms or between a dia-phragm and the tube plate (the first and the last element).

� Each element is in turn split into 6 windows (Figs. 3 and 4): twoexternal windows, the internal windows, and two cross zones.It is opportune to consider the symmetries of the unit opera-tion in order to reduce the number of variables of the resultingdynamic model. For the specific case in study, the longitudinalsymmetry can be exploited.

� There is not any vertical variation for the temperature: nodiscretization is used for the vertical axis.

Considering that the discretization is adopted also in the inletand outlet elements, the final amount of volumes is 30. Theenumeration of Fig. 3 follows the shell side process flow; the tube


Fig. 3. Horizontal half slice based on HTRI standard.

Fig. 5. Mass and fouling layers of the single tube.


side is characterized by straight flow that is segregated along thelongitudinal axis (Fig. 5).

It is reasonable to consider the shell side flow as continuouscrossflow with respect to the tubes. Actually, with diaphragmshaving windows about the 25% of the shell diameter, the heat ex-change coefficients are very close to the ones obtained in crossflowas demonstrated elsewhere [24]. The fluid velocity in the shell isassumed constant in each element andwindow. Pressure drops andheat losses to the ambient are neglected. The temperature is con-stant within the window, whereas it varies from window to win-dow and from element to element. The internal and externalfouling layers are negligible with respect to the dimensions of thetube. No radial gradients are considered.

With these assumptions, the balance that involves the inlet andoutlet streams, the heat exchanged with the metal mass (at tem-perature Tshell), and the heat exchanged between the water and theexternal layer of fouling is stated as follows:

rw$cp;w$VsdTs;idt

¼ mw$cp;w$�Ts;i�1 � Ts;i

�þ hs$As$�Tw;si � Ts;i

�� hs$Ashell$

�Ts;i � Tshell;i

�(5)

Energy balances are defined also for the metal parts of theeconomizer and the external fouling layer:

Fig. 4. Different windows for


mfoul;shell$cp;foul;shell$dTw;s;i

dt¼ 1

Rs$As$

�Tw;i � Tw;s;i

�� hs$As$

�Tw;s;i � Ts;i

�(6)

mmetal$cp;metal$dTw;i

dt¼ 1

Rt$At$

�Tw;t;i � Tw;i

�� 1Rs$As$

�Tw;i � Tw;s;i

�(7)

mfoul;tube$cp;foul;tube$dTw;t;i

dt¼ ht$At$

�Tt;i � Tw;t;i

�� 1Rt$At$

�Tw;t;i � Tw;i

�(8)

The energy balance for the molten salts tube side is as follows:

rsalt$cp;salt$VtubedTt;idt

¼ msalt$cp;salt$�Tt;kðiÞ�i � Tt;i

�� ht$At$

�Tt;i � Tw;t;i

�(9)

and the balance on the metal mass for the shell is:

mshell$cp;shell$dTshell;i

dt¼ hs$Ashell$

�Ts;i � Tshell;i

�� Uext$As$

�Ts;i � Text

�(10)

the dynamic modeling.


Table 1Parameters for the pumps.

Pumps

Name P1 P2 P3Flow curve scale (m3/h) 300 300 150Head curve scale (m) 200 200 100


where the heat losses to ambient are neglected. Uext is the overallheat exchange coefficient between the economizer and theatmosphere:

Uext ¼�1hs

� ODs

Dcoib$scoibhcoib

þ ODs

Dshell$sshellkshell

��1(11)

The Nusselt for external conditions (natural convection) is givenby Ref. [25]:

Nuair ¼

0BB@0:6þ 0:387$

ðGr$PrairÞ1=6�1þ

�0:559Prair

�9=16�8=27

1CCA

2

(12)

with Grashof number:

Gr ¼ D3s $r

2$g$b$ðT � TairÞm2

(13)

Such a detailed model for the economizer allows to predict themolten salts temperature with higher reliability. Please note thatthe overall temperature profile of molten salts is related to theeconomizer outlet temperature, for the simplified process controlscheme selected (Fig. 2). It is so detailed that the relative elongationof the tube bundle can be accurately evaluated with respect to theelongation of the shell, so as to refine also the economizer designaccounting for certain elongation peaks occurring during thetransients. For the time being, the modeling of bypass according tothe recent literature [26] is not yet implemented.

Moreover, although detailed, the object-oriented philosophyadopted to implement the resulting system allows the model toself-adapt [20] according to the operating conditions and, forinstance, the balances of the fouling are automatically removedwhenever clean conditions are detected [27]. The resultingdifferentialealgebraic equation system is solved with some veryperforming [28] and robust [29] algorithms, already implementedin commercial simulators such as PRO/II� [30], in commercialprocess optimizers such as ROMeo� [31] and well-established inthe industrial best practice [32].

Table 2Parameters for the motors.

Motors

Name M1 M2 M3Speed (rpm) 3600 3600 3600

4. Dynamic simulation of the ACSP plant

The dynamic simulation of concentrating solar plants is quitecumbersome: there are not well-established detailed models forseveral unit operations. To quote a few, the lineareparabolic solartroughs are not yet included in the common process dynamicsimulators, the mixture of molten salts are not yet well-studied andare in continuous evolution, chemicalephysical properties must bestill well-investigated, and specific ACSP units, such as the moltensalts steam generators and energy storage tanks, are significantlydifferent from the conventional units adopted elsewhere andtraditionally included in the most common model libraries.

Thus, some peculiarities of ACSP units have been implementedex novo in this work to properly reproduce the ACSP dynamicbehavior. Simulation parameters are reported in Table 1e7. Spe-cifically, the parameters for the pumps, motors and valve charac-teristics define in practice the flow passing through the solar andgeneration lines. The tuning and controller parameters are given toreproduce the process dynamics. The geometric data for the tanksallows to define the installed energy storage capacity. At last, theparameters for the solar troughs and the heat exchanger train aregiven to complete the plant configuration.

The dynamic simulation can be performed for all the transients.For the sake of conciseness, the sunrise transient only is described


in this paper. The selection of the sunrise transient is related to theparticular energy storage conditions at the sunrise. Actually, at theend of the night the hot molten salts still stored in the HOT_TANK isthe minimum of the overall day. Thus, whatever variation comingfrom the solar line induces the maximum variations also on thegeneration line since the reduced holdup in the HOT_TANK cannottotally smoothen the process dynamics. This motivation makes thesunrise transient one of the hardest for the ACSP. Moreover, at thesunrise, although the solar radiation is partially developed only, thesolar line is activated in order to harvest as much as possible thesolar radiation.

Fig. 6 shows the molten salts temperature within the genera-tion line. The highest temperature corresponds to the molten saltsoutflowing the HOT_TANK. The temperature progressively de-creases passing through the superheaters, the boiler, and theeconomizer. The temperature exiting the economizer is about290 �C and it is regulated by the PID3 that manages the water fedto the steam generation line. As it is possible to see, the temper-ature of the molten salts temporarily decreases in the HOT_TANKand the temperature of molten salts evenly decreases in thegeneration line. The temperature variation gradually reduces alongthe generation line, since PID3 preserve the final temperature onlyat 290 �C. This is crucial to prevent too low temperatures andtherefore the solidification of salts, but it has also certain relevantside effects.

While the temperature of stream S24 has no variations, inpractice, for the fast response of the control system, theremaining streams are not directly controlled. This leads to about50 �C of variation on the molten salts temperature at the inletport of molten salts of the second superheater. Since the tem-perature variation is also relatively fast, some mechanical and/orstructural problems can arise on such a unit. On the other hand,the temperature of molten salts in the economizer increases. Thisis due to the fact that the PID3 is reducing the water supply to thesteam generation to preserve to outlet temperature of 290 �C andthe economizer can recover more residual energy in the moltensalts before they leave the heat exchanger train. This aspect isclarified also considering another side effect. The trends of Fig. 7are referred to the boiler. Note that the relevant decrease of theinlet water flow (S35), which is due to the PID3 control action,induces a decrease in the water level and steam pressure withinthe boiler. As a consequence, the steam generation is temporarilylow with a reduction of the power generated with the turbine.The corresponding dynamics can be appreciated on the water/steam temperature trends (Fig. 8). Apart from these effects, thecontrol scheme well accomplishes its tasks and the responses arein good agreement with the field behavior. More elegant andmore effective control systems can be studied in the next futureto improve the operations.


Table 5Parameters for the tanks.

Thermal energy storage tanks

Name COLD_TANK HOT_TANKDiameter (m) 29 24Length (m) 13 10Heat transfer coefficient (kW/m2/K) 0.001 0.001Natural convection heat transfer

coefficient (kW/m2/K)0.1 0.001

Table 3Parameters for the valves.

Valves

Name XV1 XV2 XV5 XV6 XV7CV 500 500 300 120 500


5. Thermal energy storage

Fig. 9 reports the liquid holdup within the COLD_TANK andHOT_TANK. Trends show the linear behavior of thermal energystorage of molten salts. The liquid holdup of hot molten salts(thermal energy storage) gradually increases daytime in spite of theliquid holdup of cold molten salts. This is due to the solar linedesign, which is overdimensioned with respect to the generationline. It means that when the solar radiation is harvested, a portion issent directly to generate steam while the remaining portion iscollected as thermal energy in the HOT_TANK. Conversely, the hotholdup is consumed during the night to supply the steam genera-tion although the solar radiation is unavailable. For the ACSP, theenergy storage is related to the quantity of hot molten salts storedin the HOT_TANK. Thus, the energy stored is a fully capacitivesystem and the liquid level directly corresponds to the availablethermal energy storage, which is governed by the total mass con-servation principle:

Energy Storage ¼ Input� Outputþ Production (14)

Since no reactions occur in the energy storage tanks, for theselected energy storage the conservation principle reduces to:

Energy Storage ¼ Input� Output (15)

dMEnergy Storage

dt¼ _Min � _Mout (16)

whereM ¼ rV is the mass contained in the energy storage tank andM_ is the mass flow inflowing and outflowing the tank. The tem-perature within the energy storage tank has negligible variationsand the density r of molten salts is assumed constant:

rEnergy StoragedVEnergy Storage

dt¼ rin

_F in � rout_Fout (17)

rEnergy Storage ¼ rin ¼ rout (18)

where V is the tank volume and _F is the volumetric flow. Since theenergy storage tanks have cylindrical shape, the tank horizontalsection A is constant and can exit the derivative term:

AEnergy StoragedhEnergy Storage

dt¼ _F in � _Fout (19)

Table 4Parameters for the controllers.

Control loops

Name PID1 PID2 PID3Proportional gain 0.1 0.1 0.005Integral reset rate 0.1 0.1 0.1Controller action Reverse Reverse DirectSet point 975527 kg/h 487765 kg/h 290 �CProcess variable S5.W S9.W S24.T


where h is the molten salts liquid level. Contextualizing the balancefor the Archimede plant (with reference to the layout of Fig. 2), theenergy storage assumes the following form:

AHOT TANKdhHOT TANK

dt¼ _FS21 � _FS8 (20)

ACOLD TANKdhCOLD TANK

dt¼ _FS24 � _FS4 (21)

It is worth underlining that the energy storage technologyadopted for the ACSP is the so-called direct storage with two tanks.Briefly, the direct storage means that the energy storage is place onthe main process streams and it is directly fed by energy, withoutany utility/process/heat exchanger [3]. Using this technology,_FS21s _FS8 and _FS24s _FS4 and, therefore, the accumulations arenonzero (either negative or positive). If we assume, and thisassumption is reasonable for the simplification adopted in thesimulation, that the molten salts are incompressible and if weneglect the fast transient of filling operation of the streams ofmolten salts:

_FS8 ¼ _FS24 (22)

_FS4 ¼ _FS21 (23)

with the obvious consequence of having complementary behaviorsfor the liquid holdups of the HOT_TANK and COLD_TANK.

The energy storage linearly increases/decreases in volume ac-cording to the inflow and outflow. Specifically, during the night itincreases the level of molten salts since:

_Fcold tankin ¼ _FS24 > 0 (24)

while:

_Fcold tankout ¼ _FS4 ¼ 0 (25)

Conversely, the HOT_TANK volume increases daytime, when:

_Fhot tankin ¼ _FS21 > _F

hot tankout ¼ _FS8 > 0 (26)

The pseudo-constant periods of Fig. 6 are in correspondencewith startup/shutdown procedures of the solar line. At thebeginning and end of the day the levels of thermal energy storagetanks correspond with each other to have periodical daily opera-tions (no meteorological conditions such as clouds or dust are

Table 6Parameters for the solar trough.

Solar trough

Name PIP1Volume (m3) 20.78Heat transfer area (m2) 972


Table 7Parameters for the heat exchangers.

Heat exchanger

Name Economizer Boiler Superheater-1 Superheater-2

Heat transfer area (m2) 150 330 16 15Overall heat transfer

coefficient (kW/m2/K)1.04 1.044 0.88 0.88

Natural convection heattransfercoefficient (kW/m2/K)

0.1 0.1 0.1 0.1


considered here). Moreover, the HOT_TANK has larger variationsof the holdup level with respect to the COLD_TANK due to thedifferent diameters. Different diameters are due to safety reasons;actually, the COLD_TANK has double volume with respect to theHOT_TANK and it works as blowdown vessel for non-conventionalor emergency conditions: in fact, it can contain the whole quantityof molten salts circulating in the solar plant and stored in theHOT_TANK.

Fig. 6. Dynamics on the temp

Fig. 7. Boiler d


6. Discussion

From a general viewpoint, renewable energies are progressivelyincreasing their presence and impact in several fields such as theminimization of energy wastage in industrial processes [33], ofcarbon footprint of biomass value chains [34], of environmentalproblems [35] to quote a few. Concentrating solar surely is apromising technology, but it still has certain open operational is-sues that can make it less appealing than other rival technologies.Some key-issues are:

� The need to keep the salt molten. It means that very fasttransients may occur on the solar line and all the operationunits on this line may have shorter life cycles for strong dy-namics and frequent changes of operating conditions, asdemonstrated in this paper.

� The need to get the maximum economic benefit from the po-wer generation. This is required for fast payback and to bridgethe gap with other more energetically appealing renewabletechnologies. To do so, there is the need to optimize the hot

erature of molten salts.

ynamics.


Fig. 8. Dynamics of the temperature of water/steam.


molten salts flow supplied to the steam generation looking atthe peaks of the local/national electric demand (related tohourly forecasting) and provide the surplus of energy in cor-respondence with the peaks.

that must be coupled with more practical issues such as the well-known corrosion problems of CSP or the cleanliness of the solartroughs to quote a few.

However, tackling the former key-issues means to significantlyimprove the performances of CSP technology and its net presentvalue as well as to prolong the life cycle of CSP unit operations.

The dynamic simulation plays a relevant role in solving theseproblems since it allows to assess a priori the effectiveness of thecontrol schemes and the procedures to manage the very fasttransients on the solar line or to estimate in advance the futureperformance of the units involved in CSP according to their age andoperating conditions. Nevertheless, higher the accuracy that wewant on the assessments and previsions, deeper the level of detailsthat we need in the dynamical model for simulating the CSP.Although simplified, the dynamic simulation presented in this pa-per is a good tool to define certain guidelines and to identify thecritical points for simulating CSP dynamics. For example, there isthe need to optimize the startup (but also the shutdown) procedureto save energy and safeguard unit operations. Moreover, there is theneed to activate the solar panels only when the receiver is filled bymolten salts to prevent degradation problems or heterogeneoustemperatures within the same receivers.

Fig. 9. Dynamic behavior of energy storage tanks.


7. Conclusions and future developments

The dynamic simulation is a key-tool to improve the perfor-mances and the life-cycle of the concentrating solar plants. Asimplified dynamic simulation is useful to preliminary check thecontrol systems and procedures to perform the startup and shut-down operations to save the panels and receivers and the processunits as well, but also to identify the main critical aspects inoperating the plant. A more detailed dynamic simulation couldallow to optimize not only the process transients, but also the netincome of the plant by foreseeing the energy demand and,accordingly, storing the energy in advance and generating themaximum power when required. Detailed simulation means theintroduction of dedicated models for the heat exchanger (moltensalts e water), solar panel, and molten salts storages.

The possibility to simulate the process dynamics of a concen-trating solar plant and its related thermal energy storage opensnew appealing studies and activities for the next future, such as theintegration of concentrating solar energy with traditional chemicalprocesses, intensifying them and supplying their heat requestswith renewable energy and progressively moving toward sustain-able productions of chemical commodities. At last, the heat inte-gration of concentrating solar energy and chemical processesshould lead to the synergistic result of define, compare, assess, anddesign novel chemical energy storage technologies for solar plants.

Nomenclature

AcronymsACSP Archimede Concentrating Solar PowerM[number] MotorP[number] PumpPID[number] ControllerS[number] StreamXV[number] Valve

SymbolsA heat exchange areaB volumetric thermal expansion coefficientcp specific heatg acceleration due to Earth’s gravityH, h enthalpy, specific enthalpyhs heat exchange coefficient



m kinematic viscositym mass flowQ heat exchangedr densityR heat exchange resistancet timeT temperature of molten saltsTM temperature of metal wallU external overall heat exchange coefficientV volume

Subscriptsfoul foulingi i-th heat exchanger sectorin inletout outletrif referenceshell shell sidetube tube sidew water

References

[1] Hennings W, Markewitz P, Vogele S. Energy storage an overview of the cur-rent state of technology. BWK e Energie-Fachmagazine 2011;63(5):53e8.

[2] Wilson IAG, McGregor PG, Hall PJ. Energy storage in the UK electrical network:estimation of the scale and review of technology options. Energy Policy 2010;38(8):4099e106.

[3] Vitte P, Manenti F, Pierucci S, Joulia X, Buzzi-Ferraris G. Dynamic simulationof concentrating solar plants. Chemical Engineering Transactions 2012;29:235e40.

[4] Huang CP, T-Raissi A. Analysis of sulfur-iodine thermochemical cycle for solarhydrogen production. Part I: decomposition of sulfuric acid. Solar Energy2005;78(5):632e46.

[5] Nzihou A, Flamant G, Stanmore B. Synthetic fuels from biomass usingconcentrated solar energy e a review. Energy 2012;42(1):121e31.

[6] Piemonte V, De Falco M, Giaconia A, Tarquini P, Iaquaniello G. Life cycleassessment of a concentrated solar power plant for the production of enrichedmethane by steam reforming process. Chemical Engineering Transactions2010;21:25e30.

[7] Reddy VS, Kaushik SC, Tyagi SK. Exergetic analysis and performance evalua-tion of parabolic trough concentrating solar thermal power plant (PTCSTPP).Energy 2012;39(1):258e73.

[8] Nixon JD, Dey PK, Davies PA. Which is the best solar thermal collectiontechnology for electricity generation in north-west India? Evaluation of op-tions using the analytical hierarchy process. Energy 2010;35(12):5230e40.

[9] Nemet A, Kleme�s JJ. Optimising the temperature of heat storage to serveprocesses with varying supply and demand e captured solar energy curve.Chemical Engineering Transactions 2011;25:605e10.

[10] Quijera JA, Alriols MG, Labidi J. Usage of solar energy in an industrial process.Chemical Engineering Transactions 2011;25:875e80.

[11] Nemet A, Klemes Jj, Varbanov PS, Kravanja Z. Methodology for maximising theuse of renewables with variable availability. Energy 2012;44(1):29e37.

[12] Calise F, Palombo A, Vanoli L. Design and dynamic simulation of a novelpolygeneration system fed by vegetable oil and by solar energy. EnergyConversion and Management 2012;60:204e13.


[13] Cheng ZD, He YL, Cui FQ, Xu RJ, Tao YB. Numerical simulation of a parabolictrough solar collector with nonuniform solar flux conditions by coupling FVMand MCRT method. Solar Energy 2012;86(6):1770e84.

[14] de Arce R, Mahia R, Medina E, Escribano G. A simulation of the economicimpact of renewable energy development in Morocco. Energy Policy 2012;46:335e45.

[15] Xu E, Wang Z, Wei G, Zhuang J. Dynamic simulation of thermal energy storagesystem of Badaling 1 MW solar power tower plant. Renewable Energy 2012;39(1):455e62.

[16] Xu R, Wiesner TF. Dynamic model of a solar thermochemical water-splittingreactor with integrated energy collection and storage. International Journalof Hydrogen Energy 2012;37(3):2210e23.

[17] Powell KM, Edgar TF. Modeling and control of a solar thermal power plantwith thermal energy storage. Chemical Engineering Science 2012;71:138e45.

[18] Manenti F. Natural gas operations: considerations on process transients,design, and control. ISA Transactions 2012;51:317e24.

[19] Biegler LT, Grossmann IE. Retrospective on optimization. Computers &Chemical Engineering 2004;28(8):1169e92.

[20] Manenti F. Considerations on nonlinear model predictive control techniques.Computers & Chemical Engineering 2011;35(11):2491e509.

[21] Manenti F, Rovaglio M. Integrated multilevel optimization in large-scalepoly(ethylene terephthalate) plants. Industrial & Engineering ChemistryResearch 2008;47(1):92e104.

[22] Stephanopoulos G. Chemical process control: an introduction to theory andpractice. Englewood Cliffs, NJ: Prentice-Hall; 1984.

[23] Amato A, Compare M, Gallisto M, Maccari A, Paganelli M, Zio E. Businessinterruption and loss of assets risk assessment in support of the design of aninnovative concentrating solar power plant. Renewable Energy 2011;36(5):1558e67.

[24] Taborek J. Performance of shell-and-tube heat exchanger with double-segmental baffles. College Station, Texas: Heat Transfer Research Inc; 1972.

[25] Churchill H, Chu HS. Correlating equations for laminar and turbulent freeconvection from a horizontal cylinder. International Journal of Heat MassTransfer 1975.

[26] Manenti G. A calculation procedure for a heat exchanger and bypass equip-ment. Computers & Chemical Engineering 2011;35(11):2378e88.

[27] Dones I, Manenti F, Preisig HA, Buzzi-Ferraris G. Nonlinear model predictivecontrol: a self-adaptive approach. Industrial & Engineering ChemistryResearch 2010;49(10):4782e91.

[28] Manenti F, Dones I, Buzzi-Ferraris G, Preisig HA. Efficient numerical solver forpartially structured differential and algebraic equation systems. Industrial &Engineering Chemistry Research 2009;48(22):9979e84.

[29] Buzzi-Ferraris G, Manenti F. BzzMath: library overview and recent advances innumerical methods. Computer Aided Chemical Engineering 2012;30(2):1312e6.

[30] Signor S, Manenti F, Grottoli MG, Fabbri P, Pierucci S. Sulfur recovery units:adaptive simulation and model validation on an industrial plant. Industrial &Engineering Chemistry Research 2010;49(12):5714e24.

[31] Manenti F, Buzzi-Ferraris G, Pierucci S, Rovaglio M, Gulati H. Process dynamicoptimization using ROMeo. Computer Aided Chemical Engineering 2011;29:452e6.

[32] Manenti F, Grottoli MG, Pierucci S. Online data reconciliation with poorredundancy systems. Industrial & Engineering Chemistry Research 2011;50(24):14105e14.

[33] Kleme�s JJ, Varbanov PS, Pierucci S, Huisingh D. Minimising emissions andenergy wastage by improved industrial processes and integration of renew-able energy. Journal of Cleaner Production 2010;18(9):843e7.

[34] Lam HL, Varbanov P, Kleme�s J. Minimising carbon footprint of regionalbiomass supply chains. Resources, Conservation and Recycling 2010;54(5):303e9.

[35] Vaccari G, Tamburini E, Sgualdino G, Urbaniec K, Kleme�s J. Overview of theenvironmental problems in beet sugar processing: possible solutions. Journalof Cleaner Production 2005;13(5):499e507.


dynamic simulation of concentrating solar power plant and two-tanks direct thermal energy storage

Documents