DEVELOPMENT OF A STRUCTURED THERMOCLINE THERMAL ENERGY STORAGE SYSTEM

Transcription

1 DEVELOPMENT OF A STRUCTURED THERMOCLINE THERMAL ENERGY STORAGE SYSTEM Brad M. Brown Matt N. Straer R. Paneer Selvam Univerity of Arkana Department of Civil Engineering 4190 Bell Engineering Center Fayetteville, Arkana, USA bradmbrown001@gmail.com mtrae@uark.edu rp@uark.edu ABSTRACT Thermal energy torage (TES) ha become an area of increaed focu a part of the effort to increae the economic viability of concentrating olar power plant (CSP plant). Single-tank, thermocline TES ytem have been invetigated to lower ytem cot. Single-tank ytem reduce material cot, compared to material cot of twotank, liquid TES ytem, by toring both the hot and cold heat tranfer fluid (HTF) in the ame tank. Additionally, in thermocline ytem, the majority of the energy torage tank volume i occupied by aggregate, reducing the neceary volume of expenive HTF. Thi reduction in the neceary volume of HTF coupled with the need to only contruct one energy torage tank, make ingle-tank ytem approximately 35% cheaper than two-tank ytem (Broeau et al. 2005). The main problem encountered in thermocline, TES ytem i thermal ratcheting. Thermal ratcheting occur during thermal cycling of the TES ytem. Increaed tank temperature during the charge cycle reult in thermal expanion of the tank; expanion of the tank allow maller material from the packed aggregate bed to ettle toward it bae. When the tank cool and contract during the dicharge cycle, reidual tree are introduced to the tank wall due to the material ettlement. Repeated thermal cycle can reult in catatrophic rupture of the tank. Thi paper focue on the development of a tructured thermocline TES ytem with concrete replacing the typical, packed-bed filler material. Replacing the packed bed with tructured concrete avoid the iue of thermal ratcheting. Solar alt i incorporated a the heat tranfer fluid (HTF). Finite-difference baed numerical modeling wa utilized to optimize the geometry of the tructured concrete filler material. Modeling reult indicated that the filler concrete geometry could be optimized to allow for energy torage-to-energy retrieval efficiency of up to 65%. 1. INTRODUCTION Solar energy i a renewable energy ource that ha many poitive attribute. Two prominent attribute are that olar energy i an unlimited power ource and that olar power can be harveted in a relatively clean manner. Concentrating olar power plant focu the un emiion to heat a HTF fluid that i run into a power plant power block to drive the generation of electricity. A problem with CSP plant i that they cannot produce electricity when there i cloud cover or during the night. TES ytem are being developed to continue to upply heat to the power plant during time when direct unlight i not available. Three form of energy torage for TES ytem concept exit: enible heat, latent heat, and chemical torage. Of the three, enible heat torage ha been uggeted a the mot practical option to tore thermal energy (Herrmann and Kearny 2002). Concrete, which ha an approximate media cot of $1 per kilowatt-hour thermal (Hermann and Kearny 2002), ha been reearched a a enible heat torage medium. 1

2 Concrete and molten olar alt were incorporated in a TES ytem and proved to be a promiing combination. The molten olar alt incorporated in uch TES i capable of operating at temperature exceeding 500 C (Coatal Chemical Company 2011). The Univerity of Arkana ha developed concrete mixture that urvived teting at temperature of up to 600 C (John et al. 2010). Uing the combination of molten olar alt and concrete, TES ytem have been reearched and teted through the Univerity of Arkana. Preliminary teting incorporating concrete with an embedded, tainleteel-tube heat exchanger allowed for temperature in the concrete torage media to reach temperature of up to 450 C (Skinner et al. 2011). However, it wa concluded that the cot of the tainle teel tubing and concrete mix deign needed for thi method of TES rendered thi ytem too expenive to achieve the Department of Energy (DOE ) goal for olar, thermal energy torage by 2020: 6 cent per kilowatt-hour with hour of torage (DOE 2011). In thi paper, a thermocline-type TES ytem i propoed and developed, incorporating concrete a a tructured filler material and molten olar alt a the HTF. A thermocline TES i a ingle-tank, thermal energy torage unit; the majority of the tank volume i occupied by an aggregate or other filler material. The filler material i incorporated to reduce the neceary quantity of high cot HTF. Within the region of the tank occupied by the filler material, thermal tratification occur. Thi tratification i due to the natural buoyant force of the HTF reulting from the fluid temperature difference within the tank. Thermal tratification reult in a hot region in the upper portion of the thermocline and a cold region in the lower portion of the thermocline with a thermal gradient eparating the two region (Mawire et al. 2009). During the charging (energy torage) cycle, high temperature HTF i pumped into the top of the tank while allowing low temperature HTF to exit the bae of the tank. Pumping i conducted at a low flow rate to prevent mixing of the high and low temperature HTF. In a imilar manner, during the dicharge (energy retrieval) cycle, low temperature HTF i pumped into the bae of the tank while allowing high temperature HTF to exit the top of the tank. A in the charging phae, the low temperature HTF i pumped into the tank at a low rate to prevent mixing of the high and low temperature HTF. A major iue found in thermocline ytem i thermal ratcheting. Thermal ratcheting occur due to the ettlement of maller aggregate to the bae of the thermocline ytem during thermal cycling. A the tank wall heat up during charge cycle, they expand, allowing mall aggregate to filter down. A the tank wall cool and contract during the dicharge cycle, reidual tree are introduced to the wall due to the increaed volume of material at the bae of the tank. Over a period of time, additive tre due to repeated thermal cycle can reult in catatrophic rupture of the thermocline tank (Flueckiger et al. 2011). The iue of thermal ratcheting i avoided by incorporating a tructured filler material in place of the packed aggregate bed. Finite-difference baed numerical modeling wa implemented to optimize a tructured concrete filler material. Two unique arrangement of filler material were modeled: concrete prim with an internal, circular channel running through the center (Fig. 1A) and concrete plate tanding vertically ide by ide (Fig 1B). The exact geometrie of thee two arrangement were optimized to produce the optimum energy torage to energy retrieval ratio. 2. MODELING The two material arrangement modeled for the tructured concrete filler were concrete prim with hole running vertically along the height of the prim (axiymmetric model) and parallel concrete plate tanding vertically inide the thermocline tank (parallel plate model). Depiction of the cro ection of thermocline tank populated with each of thee two arrangement are provided in Fig. 1. Fig. 1: Axiymmetric (Left, 1A) and Parallel-Plate (Right, 1B) Concrete Arrangement in Thermocline Tank Aumption were made in the contruction of the heat tranfer model for each of the two concrete arrangement that allowed the 3D heat tranfer model to be implified to 2D model. 2.1 AXISYMMETRIC MODEL In contructing the axiymmetric model, the concrete prim were implified to concrete cylinder with hole running through their center. The model only conidered one of the prim compoing the filler material in the tank (reference Fig. 1A for tank cro ection). From the model of one prim, the total amount of energy that could be tored in the ytem and that could be retrieved from the ytem were calculated and will be dicued later. 2

3 Fig. 2 illutrate the cro ection of the axiymmetric model. Fig. 2A diplay the plane view, a would be een looking down the length of the implified prim. Labeled in thi cro ection are the radiu of the HTF flow channel through the tube (Ri) and the radiu of the implified prim (Ro). Fig. 2B diplay the plane view, a would be een if the prim wa cut along it longitudinal axi. Labeled in Fig. 2B are the length (height) of the tructured filler material (L) and the fluid flow velocity (v). The HTF velocity i aumed to be poitive to the right (correponding to a charge cycle) and negative to the left (correponding to a dicharge cycle). The axiymmetric model aume that heat i tranferred perpendicular to the direction of fluid flow and parallel to the direction of fluid flow. Fig. 2: Cro Section (Left, 2A) and Longitudinal Cro Section (Right, 2B) of Axiymmetric Model 2.2 PARALLEL-PLATE MODEL In contructing the parallel plate model, a ingle egment of thickne To wa conidered (See hatched ection of Fig. 3B). It thickne wa compoed of one-half of a concrete plate thickne and one-half of a HTF flow channel thickne; it wa aumed to be of length L. Taking thi egment a the bai for the model wa a valid approach, a all of the concrete plate in the model are uniform and paced equally. Fig. 3 illutrate the cro ection (3A) and longitudinal cro ection (3B) of the parallel-plate model. Labeled in Fig. 3B are half the ditance between the parallel plate (Ti) and the egment conidered when contructing the model (To). The HTF velocity i labeled with poitive velocity to the right (correponding to a charging cycle) and negative velocity to the left (correponding to a dicharge cycle) a in the cae of the axiymmetric model; finally the length of the filler concrete i labeled L in Fig. 3B. At the top of Fig. 3A, a length of 1 i labeled in the x direction. Thi unit width i labeled becaue the model aumed the width of the plate to be a unit of one for each increment of height (EQ.2). The HTF in contact with all of the concrete material at any pecified height i aumed to be at a uniform temperature; therefore, it i valid to aume that heat tranfer will only occur in two direction: normal to the direction of the fluid flow and parallel to the direction of fluid flow. Fig. 3: Cro Section (Left, 3A) and Longitudinal Cro Section (Right, 3B) of Parallel-Plate Model 2.3 EQUATIONS GOVERNING NUMERIC MODEL The numerical model for both concrete geometrie were contructed uing the finite difference method (FDM). The model conider conduction and forced convection a mode of heat tranfer. Two unique cenario of heat tranfer were identified in the thermocline ytem: heat tranfer at the interface between two diimilar material and heat tranfer through a uniform material (Schmidt and Willmott 1981). A thermal gradient i known to exit along a portion of the filler material length; therefore, it i neceary to model the heat diffuion through thi material to attain an accurate value for the heat energy tored in the material. The governing heat diffuion equation can be een below for the axiymmetric model (EQ.1) and the parallel-plate model (EQ.2) (Selvam 2011). The ubcript m correpond to filler material. Uing EQ.1 and EQ.2, it wa poible to model both filler material arrangement a 2D ytem. The heat tranfer from the HTF to the filler material through convection coefficient h i governed by EQ.3 (Selvam 2011). The ubcript m correpond to filler material and the ubcript f correpond to HTF. Additionally, v correpond to the velocity of the HTF, P correpond to the perimeter of the fluid-flow area cro-ection, correpond to the cro-ectional area of the fluid flow, and and correpond the HTF denity and pecific heat capacity repectively. Due to the difference in the cro ectional geometrie of the two model fluid flow channel, the convection coefficient for the two model had to be calculated eparately. The procedure ued in calculating the 3

4 convection coefficient for both the parallel plate model and the axiymmetric model i provided below (Çengel and Ghajar 2011). Uing EQ.4 and EQ.5, it wa determined that the fluid flow in both model wa laminar. The model aumed that the heat flux, from the HTF to the filler material, wa uniform around the material cro ection at any length (height) of the filler material. From thi aumption and knowing the hape of the fluid channel geometrie, value for the Nuelt number were attained and are provided below (EQ.6 for axiymmetric model and EQ.7 for parallel-plate model) (Çengel and Ghajar 2011). correpond to the hydraulic diameter of the channel; the hydraulic diameter i equivalent to four time the cro-ectional area of the flow channel divided by the perimeter of the channel cro ection. The Nuelt number for the parallel-plate model i provided a a range; thi i becaue the lower end correpond to a channel width-to-thickne ratio of eight and the upper end correpond to a ratio of infinity. The value of the Nuelt number changed with each alteration of the fluid flow channel, a thi changed the channel widthto-thickne ratio. The amount of heat tranported from the concrete-htf interface into the concrete due to conduction at any point mut equal to the amount of energy tranferred to the concrete urface at that point from the adjacent HTF through the convection coefficient h. Uing thi fact, EQ.10, EQ.1, and EQ.2 or EQ.3 made it poible to iteratively olve for the temperature of the concrete and HTF along their interface. It wa aumed that no heat tranfer occur outide the region occupied by the filler material (EQ.11). Thi i a valid aumption when conidering a thermocline ytem, becaue all HTF outide the filler material will either be in at the minimum or maximum operating temperature. During the charging cycle of a thermocline, hot HTF enter the top of the tank and cold HTF exit the bottom. EQ. 12 account for thi, etting the temperature of the HTF at the left (top) of the thermocline region equal to the hot operating temperature ( u correpond to the magnitude of V ). Uing the appropriate Nuelt number for each model, the value of the convection coefficient wa calculated uing EQ.8. Likewie, during the dicharge cycle of a thermocline, cold HTF enter the right (bae) of the tank and hot HTF exit the left (top). EQ.13 account for thi, etting the temperature of the HTF at the right (bae) of the thermocline region equal to the cold operating temperature. 2.4 BOUNDARY CONDITIONS It wa aumed that there would be no heat tranfer normal to the outer urface of each concrete element (EQ.9). The end of the element were aumed to be adiabatic becaue the ytem wa well inulated. Each filler material type wa aumed to exhibit thermal ymmetry; within each filler material configuration, each concrete element wa of the ame ize and had the ame quantity of HTF flowing through it. Therefore, the external temperature of the adjacent concrete element would be at the ame temperature, reulting in zero heat tranfer. Fig. 4 illutrate the boundary condition a they were applied to the axiymmetric model. \ Fig. 4: Axiymmetric Model Boundary Condition 4

5 2.5 MODELING RESULTS During a charging cycle, HTF at enter the thermocline tank from the left (top); the filler concrete i initially aumed to be at. During a dicharge cycle, HTF at enter the thermocline tank from the right (bae); the final temperature ditribution attained from the charging cycle i taken a the initial temperature ditribution of the concrete filler. At each time tep, EQ. 1 or EQ.2 and EQ.3 are olved implicitly; their olution are iterated until the temperature of the concrete filler and HTF at their interface converge to the ame value. The length of the time tep ued varie the number of iteration neceary for the convergence of the temperature value at each time tep. While optimizing the thermocline ytem, the following ytem parameter were held contant: the number of concrete cell (number of axiymettrical channel or parallel-plate in the energy torage tank), the length (or height) of the concrete filler material, and the maximum and minimum HTF temperature. The parameter that were varied are a follow: inner radiu/inner thickne, outer radiu/outer thickne, charging/dicharging time, and charging/dicharging HTF velocitie (See Fig. 2 for inner and outer radiu and Fig.3 for inner and outer thicknee). Many imulation were run while optimizing each model; 32 unique cae for the axiymmetric model and 20 unique cae for the parallel-plate model are reported by Brown (20ll). More information regarding the model i available from Selvam and Catro (2010). The firt model to be optimized wa the axiymmetric model. Table 1 illutrate the input parameter and the range over which their value were varied to optimize the model. TABLE 1: AXISYMMETRIC MODEL VARRIABLES Variable Inner Radiu Outer Radiu Time Velocity Number of Tube Length of the Thermocline Temperature Range Range m m m m 4 hr 6 hr m/ 0.01 m/ 1 tube 16 m 300 C C TABLE 2: PARALLEL-PLATE MODEL VARRIABLES Variable Inner Thickne Outer Thickne Time Velocity Number of Tube Length of the Thermocline Temperature Range Range m m m m 4 hr 6 hr m/ m/ 1 tube 16 m 300 C C While optimizing the model, the goal wa to maximize the dicharge efficiency of the ytem. The dicharge efficiency wa taken a the ratio of the amount of energy tored in the TES ytem to the amount of energy retrieved. For the two ytem configuration, parallel-plate and axiymmetric, the geometrically-optimized ytem providing the highet dicharge efficiencie were elected for comparion, and their pecification may be een in Table 3. TABLE 3: PARAMETER SUMMARY FOR OPTIMIZED MODELS Model Axiymmetric Parallel Plate RI or TI (m) RO or TO (m) v arge (m/) arge 5 5 ES (kwh) v i arge (m/) i arge (hr) 5 5 ER (kwh) Eff. (%) From Table 3, it i evident the parallel-plate model yielded the highet dicharge efficiency: 65.59%. After determining the optimum concrete geometrie for heat tranfer purpoe, the next area of interet wa to determine which geometry would allow for the greatet energy torage denity. A tank having a one-quare-meter cro ection wa elected for the comparion; the goal wa to ee how much energy could be retrieved from a TES ytem contructed with each of the optimized arrangement. Thi comparion i provided in Table 4. The econd model to be optimized wa the parallel-plate model. Table 2 illutrate the input parameter and the range over which thee parameter were varied to optimize the model. 5

6 TABLE 4: ENERGY RETRIEVAL PER SQUARE METER OF TANK CROSS SECTION Cae Total Area (m 2 ) ER (kwh) ER Per (kwh) Axiymmetric Model Plate model From the reult preented in Table 4, it i evident that the parallel-plate model yielded a higher energy torage denity then the axiymmetric model did. Therefore, incorporating the parallel plate filler concrete would allow for the contruction of a maller, therefore cheaper, torage tank then would be required to tore and retrieve the ame quantity of energy uing the axiymmetric filler material arrangement. The final comparion of the two thermocline filler material geometrie wa of their thermocline zone hape. The thermocline zone conit of a temperature gradient eparating a high-temperature upper region and a lowtemperature lower region; it preence i crucial for the ucceful operation of a thermocline TES ytem. If the thermal gradient diappear from the tank or occupie an exceive length (height) of the tank, ytem efficiency will be decreaed. Fig. 5 and Fig. 6 diplay the charging cycle of thermocline tank populated with filler material of the axiymmetric and parallel plate geometry repectively. The temperature in the tank increae a hot HTF i added to the right of the tank (top) and cold HTF i removed from the left (bottom). Fig. 6: Parallel-Plate Concrete Model Charging Cycle Following the 5-hour charging cycle, the filler material having parallel-plate geometry wa found to be at the maximum operating temperature at a ditance of approximately 5 m from the HTF inlet. In comparion the filler material having axiymmetric geometry wa only heated to the maximum temperature at the ditance of approximately 3 m from the HTF inlet. Both model maintained ignificant thermal tratification throughout their charging cycle. Thi wa indicated by the hape of their temperature ditribution plot within the thermocline. The temperature ditribution plot of the two ytem indicated that a greater portion of the left-mot (upper) region of the parallel-plate model thermocline reached the maximum HTF temperature. Therefore, it wa concluded that thi model provided uperior heat tranfer characteritic then did the axiymmetric model. Fig. 7 and Fig. 8 diplay the dicharge cycle of thermocline tank populated with axiymmetric and parallel-plate concrete geometry repectively. Following the 5-hour dicharge cycle, the parallel-plate geometry material wa found to primarily be in the temperature range of 465 C to 300 C. In comparion, the axiymmetric material wa found to be in the 490 C to 300 C range. Once again, thi indicated that the parallel-plate arrangement of the filler concrete material wa more conducive to heat tranfer than wa the axiymmetric arrangement. Fig. 5: Axiymmetric Concrete Model Charging Cycle 6

7 efficiency and allowed for the mot energy retrieval per unit cro ection of the thermocline tank. 5. FUTURE WORK Moving forward from thi reearch, more work could be done to further demontrate the viability of a tructured concrete thermocline a a TES for a CSP. One area that could be invetigated further i the modeling the concrete filler arrangement between different temperature limit; thi work only conidered the range of 300 C to 585 C. Different temperature bound could lead to higher dicharge efficiencie. Additionally, concrete filler arrangement other than the two invetigated in thi work could be modeled. Fig. 7: Axiymmetric Concrete Model Dicharge Cycle Latly, a cot evaluation could be done be done to cale the viability of the tructured thermocline TES ytem veru the two-tank or packed bed ytem already in ue. 6. NOMENCLATURE v v hydraulic diameter of flow path v Nuelt number i i Fig. 8: Parallel-Plate Concrete Model Dicharge Cycle 4. CONCLUSIONS To contruct a tructured thermocline TES ytem, the concrete filler geometry hould be optimized to allow for optimum dicharge efficiency. Additionally, optimization hould lead to a filler concrete geometry that allow for the maximum quantity of energy retrieval per unit of torage tank cro-ectional area. magnitude of fluid flow velocity V HTF kinematic vicoity V ~ HTF velocity of Of the two filler material layout invetigated it wa found that the parallel-plate model provided the greatet dicharge 7

8 7. ACKNOWLEDGMENTS The reearch work wa upported by a grant from the U.S. Department of Energy (Grant # DE-FG36-08G018147) through the Univerity of Arkana to perform thi work. The opinion expreed in the paper do not reflect thoe of the reearch ponor. 8. REFERENCES Brown, B. (2011). Development of a Structured Concrete Thermocline Thermal Energy Storage Sytem. Mater Thei, Univerity of Arkana, Department of Civil Engineering. Broeau, D., Kelton, J., Ray, D., Edgar, M., Chiman, K., Emm, B. (2005). Teting of Thermocline Filler Material and Molten-Salt Heat Tranfer Fluid for Thermal Energy Storage Sytem in Parabolic Trough Power Plant. Journal of Solar Energy Engineering, vol. 127, Çengel, Y and Ghajar, A. (2011). Heat and Ma Tranfer: Fundamental and Application (4 th ed.). New York: McGraw-Hill. Coatal Chemical Company L.L.C. (2011). Hitec Solar Salt. 2011, Hitec%20Solar%20Salt.pdf DOE (2011). Thermal Storage Reearch and Development. Solar Energy Technologie Program, U.S. Department of Energy, _rnd.html Flueckiger, S., Yang, Z., Garimell, S. (2011). An integrated thermal and mechanical invetigation of moltenalt thermocline energy torage. Applied Energy xx, xxx-xxx. In Pre. Herrmann, U., Kearny, D. W. (2002). Survey of Thermal Energy Storage for Parabolic Trough Power Plant. Journal of Solar Energy Engineering, vol. 124, John, E., Hale, W., and Selvam, R. (2010). Effect of High Temperature and Heating Rate on High Strength Concrete for Ue a Thermal Energy Storage. ASME Conf.Proc., 2010(43956), Mawire, A., McPheron, M., van den Heetkamp, R., Mlatho, S. (2009). Simulated Performance of Storage Material for Pebble Bed Thermal Energy Storage (TES) Sytem. Applied Energy 86, Schmidt, F. and Willmott, A. (1981). Thermal Energy Storage and Regeneration. New York: McGraw- Hill. Selvam, R. P. (2011). Thermal Finite Difference Model. Private Communication. Selvam, R.P. and Catro, M. (2010). 3D FEM Model to Improve the Heat Tranfer in Concrete for Thermal Energy Storage in Solar Power Generation. Paper no: ES , Proceeding of ASME 2010, 4th International Conference on Energy Sutainability ES2010, May 17-22, Phoenix, AZ Skinner, J., Brown, J., and Selvam, R.P. (2011). Teting of High Performance Concrete a a Thermal Energy Storage Medium at High Temperature. Proceeding of ASME th international Conference on Energy Sutainability, ESFuelCell , Aug. 7-10, Wahington, DC 8