PRESSURE DROP ANALYSIS OF STEAM GENERATION PARABOLIC TROUGH PLANTS

PRESSURE DROP ANALYSIS OF STEAM GENERATION PARABOLIC TROUGH PLANTS Heko Schenk 1 and Tobas Hrsch 2 1 M.Sc., Researcher, Insttute of Techncal Thermodynamcs, German Aerospace Center (DLR), Stuttgart (Germany), phone: +49 (0)711 6862 740, E-Mal: heko.schenk@dlr.de 2 Dr.-Ing., Project Manager, Insttute of Techncal Thermodynamcs, German Aerospace Center (DLR), Stuttgart (Germany) Abstract Drect steam generaton (DSG) n parabolc trough plants has the potental to be more cost-effectve than ol based systems. The produced steam can drectly be used for process heat applcatons. Wthn the P3 project such a DSG plant of 108 m 2 of Soltem s PTC1800 has been nstalled on the roof of the buldng of the company Alanod. In the present paper, a two-phase pressure drop analyss of typcal steam generaton plants wth dfferent layouts s presented. A statc model has been developed for ths purpose. Parametrc studes have shown that, due to the elevated pressure drops, the length of a collector loop wthn a DSG plant s lmted. Consequently, n bgger plants collector loops need to be nstalled n parallel. In each collector loop so-called Lednegg nstabltes occur. When collector loops are nstalled n parallel, there are addtonal flow nstabltes. Those parallel flow nstabltes occure due to shadng of one or several collector loops. Changng pressure drops lead to an unfavourable reallocaton of mass flows. In the worst case, steam s superheated and the collectors get damaged due to hgh temperatures. Lednegg and parallel flow nstabltes can both be reduced by nstallng an addtonal flow resstor at the collector loop s nlet. 1. Introducton Drect steam generaton (DSG) n parabolc trough plants represents a cost-effectve alternatve to conventonal ol based or pressurzed water based systems. The produced steam can ether be used drectly n ndustral processes or to generate electrcty. The present study concentrates on process heat applcatons. Wthn the German project SOLDI the feasblty of drect process steam generaton was demonstrated n the SOPRAN test faclty at DLR Cologne [1]. Currently, n the project P3, a parabolc trough collector feld s nstalled on the roof of a buldng of the German company Alanod, [2]. The feld conssts of 108 m² of PTC 1800 collectors developed and produced by Soltem. Saturated steam s produced and fed nto a steam network. Recent works by varous authors show that flow nstablty problems wll be faced also n solar applcatons. On the one hand, the typcal Lednegg nstablty n a sngle loop mght occur, but also parallel flow dstrbuton problems may occur [1, 3]. Ths paper presents stablty analyzes for a process heat applcaton n the pressure range of 5 to 15 bar. 2. Statc modellng of the system In order to examne pressure drops occurrng n drect steam generaton plants, a Matlab model has been developed. The model s able to calculate two-phase pressure drops under steady-state condtons. In the followng sectons the layout of the model s descrbed. 2.1 Model of the drect steam generaton plant An exemplary steam generaton nstallaton s shown n fg. 1. Ths nstallaton s smlar to the layout of the P3 project. Saturated steam at a pressure of 5 bar (T sat = 151.8 C) s produced. Lqud water enters the solar feld (0) gettng evaporated by the concentrated solar rradaton. Wet steam wth the steam mass fracton x n

leaves the solar feld. In the steam drum water and steam get separated. Saturated steam leaves the steam drum, supplyng the steam network, poston s. The outlet pressure p s s predetermned by the network. Separated water leavng the steam drum s mxed wth feed water (f) and recrculated to the solar feld va a pump, poston z. The steam mass fracton at the outlet of the feld s controlled by means of the recrculaton mass flow. Fg. 1. Exemplary drect steam generaton plant 2.2 Model of the collector loop The solar feld conssts of 1 or more loops of several of Soltem s PTC 1800 collectors. One module has an aperture length of 5 meters and an aperture wdth of 1.8 m. Usually, 6 collector modules wth a total length of 30 m are mounted on one trackng system. Parameters of the PTC 1800 collectors are gven n table1. aperture wdth module length absorber tube nner dameter d 1.8 m 5 m 0.0356 m focal length temperature range optcal peak effcency 0.78 m 150-250 C 68 % Table 1. Collector specfcatons Fg. 2. Collector loop model Fg. 2 shows a scheme of the collector loop model. There are n collectors wth a length of 30 m. When several collectors are mounted n seres (n > 1), flexble ppes are n between two collectors. In the model, these are represented as four 90 -elbows wth 5 straght ppes n between.

Pressure drops n the elbows are calculated wth a correlaton, proposed by Chsholm [4]. Ths pressure drop depends on the flud s propertes, mass flow and the rato of radus of the bend and nner ppe dameter, R/d, Δ p f m, d, R, p, x. In the model, straght ppes are subdvded axally nto control volumes. For the collector model, a dscretsaton of 15 elements n a 30 m collector has been used. The accuracy of the results cannot be sgnfcantly ncreased wth a fner dscretsaton. Fg. 2 on the rght shows element of the dscretsaton. The flud enters the element wth the propertes of element -1. The flud n the element has propertes. The pressure drop n one control volume s calculated wth a correlaton proposed by Fredel [5]: dp dz p p 1 f m, d, x, p dz where d represents the tube dameter of the element. One has to acknowledge that the Fredel correlaton was developed for smooth ppes. Hence, pressure drops tend to underestmate, snce absorber tubes can get ncrusted. For the dmensonng of a collector loop, pressure losses should be multpled wth an approprate safety factor. By observng the control volume an equaton for the flud s enthalpes and the heat transferred nto the element can be wrtten as follows: h f DNI IAM,, T, T A Q m h, 1 eff, 0 a,where DNI eff represents the effectve drect normal rradaton, IAM the ncdent angle modfer, 0 the optcal peak effcency, and A the aperture area of the element.. Heat losses of the absorber tube depend on the dfference between flud and ambent temperature (T and T a ). Optcal effcency and thermal losses are calculated wth emprcal correlatons whch are based on measurements n the DLR n Cologne [6]. Steam mass fractons at the nlet and the outlet depend on the flud s enthalpy and pressure: p, h x f p h x f, 1 1 1 3. Pressure drop as a functon of a sngle collector loop s length and ts outlet pressure In the P3 project, a total of 60 m absorber tube s mounted n seres. Possble layouts for further projects where more steam s to be produced have to be nvestgated. Hence, a frst step s to nvestgate the mpact of the loop length on the performance. For ths purpose, the layout shown n fg. 1 s examned. For the smulatons n ths secton the followng hypotheses are appled Effectve solar rradance, pressure and steam mass fracton at the solar feld outlet are parameters The saturated steam mass flow leavng the steam drum s always compensated by the feed water mass flow. The solar feld conssts of one collector loop, composed of several collectors wth 30 m length each. The number of collectors n the loop s vared n order to nvestgate the pressure losses and the maxmal temperatures n the absorber tube. Table shows 2 the nput values for the parameter study. Three dfferent outlet pressures, 5, 10, and 15 bar are examned. Furthermore, for each pressure level three representatve steam mass fractons are taken nto account. p n T a T f DNI eff x n 5 bar, 10 bar, 15 bar 20 C 100 C 1000 W/m² 0.1, 0.15, 0.2 Table 2. Parameters for the maxmal absorber tube length study Results are shown n fg. 3. In the left dagram total pressure losses as a functon of the collector loop s length are plotted. The dagram on the rght gves the maxmum temperature n the absorber tube. One can observe that, n ths constellaton, generally a lower steam mass fracton leads to hgher pressure losses, snce

the recrculaton mass flow s ncreased. A further concluson s that hgher outlet pressures reduce the absolute pressure drop. Obvously, hgher outlet pressures ental hgher saturaton temperatures As a thrd concluson, the one mght remark that wth 5 bar outlet pressure the maxmal temperature n the collector ncreases extremely wth the collector length. The outlet temperature s always 151,8 C due to the saturaton pressure of 5 bar. However, wth an absorber tube length of 210 m the maxmum temperature would 175 C, occurrng near the collector loop s nlet where the saturaton temperature s hgher. Ths leads to hgher heat losses. By ncreasng the absorber tube dameter pressure losses could be reduced but, on the other hand, heat losses rse due to larger heat transfer area. As a consequence, t s favourable not to exceed a certan collector length, but to mount several loops n parallel. Fg. 3. Pressure drop and max. temperature as a functon of the absorber tube length 4. Instable pressure drop behavour 4.1 Lednegg nstabltes n a sngle collector loop A flow nstablty n a sngle collector loop occurs f an ncrease of mass flow leads to a decrease n pressure drop. The normally contnuously rsng pressure-drop-mass-flow curve then shows a local maxmum and an mposed pressure drop on a ppe system may yeld two dfferent mass flows. Under certan crcumstances, the system can jump between these two states. Ths so-called Lednegg nstablty [7] occurs especally n two-phase flow condtons, where the pressure drop s sgnfcantly nfluenced by the steam fracton. Also the solar steam generaton can suffer from ths problem as shown n the followng. The Lednegg nstablty s analyzed usng a sngle reference collector loop extracted from the solar feld system. It conssts of 2 collectors of 30 m each mounted n seres. Under constant ambent condtons the nlet mass flow and the water nlet temperatures are vared resultng n a wde range of outlet steam condtons. Outlet pressure p n remans constant. Input parameters are lsted n table 3. T 0 p n T n T a DNI eff solar feld layout 10-150 C 5 bar 151.8 (Tsat) 20 C 1000 W/m² 1 loop of 2*30 m Table 3. Parameters of the sngle loop pressure drop nstablty study The results of the smulaton are shown n fg 4 on the left sde. Wth an nlet temperature T 0 of 150 C, pressure drops ncrease wth the augmentng mass flow. Lower nlet water temperatures lead to an unsteady pressure drop curve. From the left dagram n fg. 4, one reckons that 2 flows correspond to the same pressure drop. In combnaton wth the flud pumps characterstc the mass flow may turn out to be not controllable.

Ths s the case when a plus n pump power does not lead to an ncrease n mass flow. If the system falls nto a confguraton wth small mass flow the steam s superheated n the collectors. When a certan temperature s exceeded collectors could be damaged. It can be clearly seen that crtcal stuatons occur at low nlet temperatures (at the gven pressure). In a real system lke the one n fg. 1 these temperatures can be reached f a small recrculaton rate goes along wth low feed water temperatures enterng the system. Specal attenton has to be pad for the mornng hours where the temperatures n the solar system are stll low. The desgn and control system of the plant have to consder ths aspect. Fg. 4. Impact of a flow resstance on the pressure drop characterstcs left: wthout flow resstance rght: wth flow resstance of j=6 In once-through bolers of coal fred plants smlar problems occur. For ths reason, durng start up mass flows and heat are controlled n such a manner, that nstable pressure drops are avoded, [8]. In solar thermal power plants, boundary condtons such as ambent temperature and rradaton are not stable and cannot be nfluenced. An approprate method to avod pressure drop fluctuatons s the nstallaton of an addtonal flow resstor at the nlet (sngle phase flow) of the collector loop, as proposed n [1] and [9]. The pressure drop of the flow resstor s added to the pressure drop of the collector loop. Thus, a steadly ncreasng pressure drop characterstcs s obtaned. Flow resstors show a nearly parabolc pressure drop characterstcs descrbed wth the equaton. ρ 2 Δp ζ 2 u 2 where descrbes the densty of the flud. u 2 refers to the velocty of flow n m/s at the outlet and descrbes the resstance coeffcent. The flow resstor s nstalled at the collector loop s nlet where water s lqud. Its densty s consdered as constant and ts flow velocty at the outlet equals the one at the nlet of the flow resstor, u 2 = u 1 = u. Pressure drops can also be expressed as a functon of the mass flow rate multpled wth the resstance coeffcent j. Δp j m 2 8 mt j ζ 4 ρ d A parametrc study has shown that for the consdered nstallaton values, pressure drop curves of the collector loop are strctly ncreasng for j > 6, see fg. 4. 4.2 Pressure drop nstabltes n two parallel collector loops As already outlned n secton 3 a certan loop length should not be exceeded. Therefore, n ths secton a system of parallel collector loops s modelled, see fg. 5. These collectors are mounted n parallel. They are suppled by one common pump. π 2

Fg. 5. Parallel collector loops In the frst study, a system of 2 parallel collector loops wthout hydraulc resstance at the nlet s examned. As n the prevous secton nput parameters can be chosen ndependently. Inlet temperature T 0, total mass flow rate (sum for the two loops) and outlet pressure p n s held constant. Snce both loops are suppled wth one pump ther pressure drops have to be equal. The same mass flow wll crculated through both loops provded that ambent condtons are exactly the same for both. However, t occurs that one collector s shaded whle the other stll s rradated. Dfferent steam mass fractons nfluence pressure drops and the mass flows gets redstrbuted. Fg. 6 shows such a scenaro. In the upper dagram, pressure losses of a sngle collector loop and pressure losses of two collector loops n parallel are shown (double mass flow for the same pressure drop). The pressure drop curve of one sngle shaded collector s plotted, as well. As a thrd confguraton, the pressure drop of two loops n parallel wth one of them shaded s plotted. Parameters are lsted n table 4. The pressure drop of the two-loop system wth one loop shaded s very low compared to the two-loop system wth both loops rradated. Ths ndcates a strong mass flow maldstrbuton wth a hgh mass flow n the shaded and a small mass flow n the heated loop. The mass flow dstrbuton can be derved from the dagram n the followng way. We suppose that the effectve DNI s 1000 W/m² for both loops (a and b) and the ntal mass flow s 0.3 kg/s for each collector and hence 0.6 kg/s for both (c). A steam mass fracton of 0.1 s obtaned. If one collector s shaded and the mass flow s held constant pressure drops decrease from almost 0.4 bar to 0.06 bar (d). We obtan the new mass flow of the rradated collector n pont e, 0.03 kg/s. The mass flow of the shaded collector s thus 0.57 kg/s. In the dagram below the new steam mass fracton of the rradated collector of 0.98 can be read. 1 loop rradated 2 loops: Both rradated 1 loop shaded e b c d 2 loops:1 rradated, 1 shaded f a Fg. 6. Parallel flow nstabltes: 2 parallel loops wthout flow resstor at the nlet

It gets obvous that n ths constellaton the solar feld s not optmally desgned. Shaded collectors are flown through by more water than rradated ones. Due to the low mass flow n the rradated collectors, the steam mass fracton s close to 1. In the above example, f the orgnal mass flow was slghtly lower than 0.6 kg, superheated steam would be produced n the rradated loop. In the worst case, the collector could be damaged when the maxmum temperature s exceeded. The nhomogeneous mass flow repartton would be exacerbated f 3 or more collectors were nstalled n parallel. T n X 0 p n T a DNI eff rradated DNI eff shaded 150 C 0 5 bar 20 C 1000 W/m² 0 W/m² Table 4. Parameters for the parallel flow study In order to cope wth the unfavourable mass flow repartton, the smplest soluton s to add a flow resstor at the nlet of each loop. The pressure drop dfference between an rradated and a shaded collector s then lower. The two-phase pressure losses then only account for a part of the pressure losses of one loop. In a second parameter study 10 loops of 2 x 30 m are mounted n parallel n the collector feld. At each row nlet, a flow resstance wth the resstance parameter j = 6 s nstalled. On all accounts such a flow restrctor s requred n order to cope wth Lednegg nstabltes as dscussed n the last secton. Boundary condtons are the same as n the study above, see table 4. In order to choose an extremely unfavourable case, ntal steam mass fracton shall be 0.2 whch leads to a mass flow of 0.21 kg/s for each loop, see fg. 7. If 9 out of 10 collectors are shaded (bold grey lne), applyng the same method as n fg. 6, we obtan a new mass flow of 0.14 kg n the rradated collector loop. The steam mass fracton s 0.27 whch s acceptable. Superheatng of steam wll not occur. 1 loop rradated 1 loop shaded 10 loops: Irradated 10 loops:1 rradated, 9 shaded Fg. 7. Parallel flow nstabltes: 10 parallel loops wth flow resstor at each nlet The study above shows, that wth the nstallaton of a smple flow resstor at the loops nlet the mpact of parallel flow nstabltes can be reduced. Even f several collectors are shaded, steam wll not be superheated n rradated loops. Wth the smulaton tool developed the necessary resstance value of the flow restrctor can be determned. Nevertheless, the mass flow repartton n the collector loops s not optmal. When collectors are fully shaded, water s stll crculated. Heat and pressure losses occur, reducng the overall performance of the solar feld. In order to cope wth ths problem, controlled valves could be nstalled at the nlet of each loop. When collectors are shaded, water mass flow could be reduced n the respectve loops. Due to lower pressure losses the crculaton pump s electrcty consumpton s reduced. Contnuatve economc studes have to show f such an nstallaton of controlled valves s justfed from an economc pont of vew.

5. Concluson A statc Matlab model of a drect steam generatng parabolc trough plant has been developed. The man purpose of the model s to represent two-phase pressure losses. A smulaton of a typcal solar plant for process heat applcatons has shown that the length of a sngle collector loop s lmted due to pressure losses. Saturaton pressure at the loop s nlet s hgher than the pressure at the outlet. As a result, the maxmum steam temperature s attaned near the nlet of the loop. Ths behavour entals more elevated heat losses and hence a lower collector effcency. Therefore, t s wse to nstall collector loops n a hydraulcally parallel algnment. More detaled examnatons of sngle collector loops reveal that wth low nlet temperatures, so-called Lednegg nstabltes occur. These can be avoded by nstallng a flow resstor wth a parabolc pressure drop characterstcs at the loop s nlet. The pressure drop of the flow resstor s added to the two-phase pressure drop n the absorber tubes and a steadly ncreasng pressure drop characterstcs curve can be obtaned. Smulatons of parallel collector loops show that, due to shadng, parallel flow nstabltes occur. The redstrbuton s counterproductve snce rradated loops turn out wth less mass flow. Flow resstors requred for Lednegg compensaton help to reduce ths effect to a tolerable extend. Though, an deal flow dstrbuton can only be obtaned by nstallng flow control valves at the entrance to each collector loop. Acknowledgements The authors would lke to thank the German Mnstry for the Envronment, Nature Conservaton and Nuclear Safety (BMU) for the fnancal support gven to the P3 project (contract No 0329609A). References [1] M. Eck, M. Eckhoff, D. R. Krüger, J. Schöneberger, T. Hrsch, 2007, SOLDI - Unterstützende F&E zur beschleungten Marktenführung der solaren Drektverdampfung [2] T. Hrsch, K. Hennecke, D. R. Krüger, A. Lokurlu, M. Walder, The P3 Demonstraton Plant: Drect Steam Generaton for Process Heat Applcatons, 11th Solar Paces Internatonal Symposum, 2008 03 03 2008 03 07, Las Vegas (USA). [3] Y. Tatel, S. Natan, D. Barnea, Drect steam generaton n parallel ppes, Internatonal Journal of Multphase Flow 29 (2003) 1669 1683 [4] D. Chsholm, 1980, Two-Phase Flow n Bends, Internatonal Journal Multphase Flow, Vol 6, pp. 363-367 [5] L. Fredel, 1975, Modellgesetz für den Rebungsdruckverlust n der Zwephasenströmung. VDI Forschungsheft, Nr. 572, VDI-Verlag, Düsseldorf. [6] N. Janotte, S. Meser, D. Krüger, R. Ptz-Pahl, S. Fscher, H. Müller-Stenhagen, M. Walder, Bestmmung der thermschen Lestungsfähgket des Parabolrnnenkollektors PTC1800, 19. Symposum thermsche Solarenerge, Bad Staffelsten, 2009 05 06 2009 05 08 [7] M. Lednegg, 1938, Unstabltät der Strömung be natürlchem und Zwangsumlauf. De Wärme, Band 61, S. 891-898. [8] R. Doležal,1962, Durchlaufkessel, Vulkan-Verlag Dr. W. Classen, Essen [9] J. D. Pye, G. L. Morrson and M. Behna, Unsteady effects n drect steam generaton n the CLFR, ANZSES Solar Conference, 2007 10 02 2007 10 06, Alce Sprngs (Australa)