Developing a cost effective rock bed thermal energy storage system: Design and modelling

Transcription

1 Developing a cost effective rock bed thermal energy storage system: Design and modelling Hendrik Frederik Laubscher, Theodor Willem von Backström, and Frank Dinter Citation: AIP Conference Proceedings 1850, (2017); doi: / View online: View Table of Contents: Published by the American Institute of Physics Articles you may be interested in Rock bed thermal storage: Concepts and costs AIP Conference Proceedings 1734, (2016); / Parametric analysis of a packed bed thermal energy storage system AIP Conference Proceedings 1850, (2017); / Capital cost expenditure of high temperature latent and sensible thermal energy storage systems AIP Conference Proceedings 1850, (2017); / Experimental and numerical investigation of a packed-bed thermal energy storage device AIP Conference Proceedings 1850, (2017); / First operational results of a high temperature energy storage with packed bed and integration potential in CSP plants AIP Conference Proceedings 1850, (2017); / Rock-bed thermocline storage: A numerical analysis of granular bed behavior and interaction with storage tank AIP Conference Proceedings 1850, (2017); /

2 Developing a Cost Effective Rock Bed Thermal Energy Storage System: Design and Modelling Hendrik Frederik Laubscher 1, 2, a), Theodor Willem von Backström 1, 2, b) and Frank 1, 2, 3, c) Dinter 1 Department of Mechanical and Mechatronic Engineering, University of Stellenbosch, Private bag X1, Matieland 7602, South Africa, Phone: Solar Thermal Energy Research Group (STERG) 3 Prof. Dr.Ing, Eskom Chair in CSP a) Corresponding author: @sun.ac.za b) twvb@sun.ac.za c) frankdinter@sun.ac.za Abstract. Thermal energy storage is an integral part of the drive for low cost of concentrated solar power (CSP). Storage of thermal energy enables CSP plants to provide base load power. Alternative, cheaper concepts for storing thermal energy have been conceptually proposed in previous studies. Using rocks as a storage medium and air as a heat transfer fluid, the proposed concept offers the potential of lower cost storage because of the abundance and affordability of rocks. A packed rock bed thermal energy storage (TES) concept is investigated and a design for an experimental rig is done. This paper describes the design and modelling of an experimental test facility for a cost effective packed rock bed thermal energy storage system. Cost effective, simplified designs for the different subsystems of an experimental setup are developed based on the availability of materials and equipment. Modelling of this design to predict the thermal performance of the TES system is covered in this study. If the concept under consideration proves to be successful, a design that is scalable and commercially viable can be proposed for further development of an industrial thermal energy storage system. INTRODUCTION Thermal energy storage (TES) is an integral part in the drive for low cost of concentrated solar power (CSP). The developing of a cost effective TES system is covered in this study with the focus on design and the modelling of a future experimental setup. The heat transfer fluid (HTF) is air and the storage material is a packed bed of rocks. A 2.5 MWh th TES system consisting of a packed bed of rocks with a conical shape piled above the ground is being developed and construction of the experimental setup is currently underway. Figure 1 shows the schematic layout of the future experimental setup. The TES concept under investigation is designed for operation at a temperature range of up to at least 600 C. TES with a packed bed of rocks is well suited to the environment in South Africa, since the availability of the suitable rock type. Dolerite, a rock suitable for thermal storage at these temperatures, is abundantly available in the Northern Cape province of South Africa, which coincides with one of the highest solar resource spots in the world. This is an advantage because transportation costs of the storage media will be minimized if the storage facility is located close to the source the rocks. The goal of decreasing the costs of thermal storage in CSP is to decrease the levelized cost of electricity (LCOE) and to enable the supply of electricity to meet demand. The concept developed in this study was proposed by Gauché [1]. A distinctive characteristic of the concept is that the bulk TES medium is not insulated, as is the case for conventional high temperature TES. This is an inherent advantage of the poor heat transfer between individual particles in the packed rock bed. Since the storage segment in a power plant can contribute up to 11% of the LCOE, lower-cost TES methods can have a significant influence in the cost at which electricity can be produced [2]. SolarPACES 2016 AIP Conf. Proc. 1850, ; doi: / Published by AIP Publishing /$

3 FIGURE 1. Schematic of the packed bed layout section showing airflow direction and hot spot in the center DESIGN OF AN EXPERIMENTAL SETUP The experimental design and system layout is simplified and optimized to be cost effective by utilizing the available resources to construct a test rig. Using local materials as far as possible and also using standard materials that are available off the shelf, resulting in an effective method of design for low cost manufacturing. A special design is developed to accommodate the high temperature regions of the system by making use of materials that can withstand the high temperature operating conditions. Operation strategies are set in place to ensure that the outer region of the rock pile does not exceed the maximum temperature as specified in the design parameters in Table 1. Heating would be controlled by measuring the temperature at the free surface of the rocks to prevent the outlet air temperature to increase above the specified threshold. Modelling of the heat transfer characteristics such as the temperature profile in the storage medium is required in the design for the placement of the temperature and pressure measuring instrumentation. Refer to the section on CFD simulation for a qualitative distribution of the temperature profile in the TES medium. In the CSP application, the heat source of the heating fluid would be an air receiver that receives a concentrated solar flux to heat up the heat transfer fluid. A gas burner is specified in this design to provide the required heat, since a solar air receiver that can produce hot air at a temperature of up to 600 C is not available for this project. A drawing of the experimental design is illustrated in Fig. 2 showing the layout of the system and displaying a cutout view of the design. The location of the center ducting, containment structure, and the pile of rocks is described by the graphics in the drawing. An air gap between the roof canvas and the free surface of the rocks is an important design aspect and is also visible in the drawing. Design Parameters of the System TABLE 1. Design parameters of the experimental system, rated values Design parameter Value (range) Heating capacity 400 kw th Charging time (5 8) hours Total Energy Storage 1.5 MWh th MWh th Fan power 5.5 kw e Rocks weight kg Max charging temperature 550 C 600 C Max outlet temperature 40 C 60 C

4 FIGURE 2. Drawing of the experimental design showing a section view Design Considerations Material specifications for all the different components of the containment structure of the rocks and also the air ducting are specified according to the design parameters of the system. Keeping in mind that the design should be airtight, and to conform to the elevated temperatures, sealing interfaces and joints are challenging. The whole experimental setup would be situated outdoors and would be open to the elements of nature; therefore the structure should be watertight or water resistant and also resistant to UV rays. In this low cost design, simplifications for the containment structure are made to ensure that the support structure material is minimized. Since a containment structure is optional in the concept, it is not necessary to have an expensive roof structure. In this specific experimental design, it is required to have a containment structure that is airtight to accommodate the discharging process. Instead of sucking the hot air (at approximately 550 C) out of the center ducting with a blower, the blower is used to create a positive pressure in the outer region of the rocks pile underneath the roof. This way, a normal blower can be used for the charging and discharging cycles of the system and still be situated in the cold side ducting. See Fig. 3 (a) and (b) for the layout of where the fan is situated in the system. It is however possible to have a blower on the hot side of the air ducting, but it would be much too expensive for the application of an experimental setup. Stacking the rocks at the natural angle of repose has the advantage that the storage medium is self-supportive and does not require any heavy support structure to be contained. The storage media is also not constrained to a fixed geometry and can physically be cast into any geometry to adhere to the design specification or to the immediate environment. No major earthmoving is thus required before the installation of the rocks. Only the center ducting that needs to be installed vertically and level relative to the ground require preparations and ground work. Materials Specification All steel structures that are in direct contact with high temperature air are required to be manufactured of stainless steel. The cheapest stainless steel for high temperature application that also has the required strength at high temperatures is found to be of the 400 series stainless steel [3]. This is also the available stainless steel that meets the requirements and can be procured from the local steel suppliers. Insulation for the hot air ducting consists of dense mineral wool and is cladded with sheet metal to provide the necessary protection against the elements. The rocks used for the experimental setup would be 53 mm pebble diameter railway ballast that is available at the local quarry

5 Operating Conditions FIGURE 3 (a). Operation flow diagram of the experimental setup; Charging Cycle FIGURE 3 (b). Operation flow diagram of the experimental setup; Discharging Cycle ANALYTICAL HEAT TRANSFER CALCULATIONS Heat transfer and flow calculations in a packed bed of rocks have been investigated and studied in previous work done by Allen et al [2]. The Effectiveness-NTU (E-NTU) method as developed by Hughes [4], Duffie and Beckmann [5], proof to be a simple and effective method of predicting a one dimensional temperature profile. This is a numerical approximation of the Schumann equations [6]. The simplified Nusselt number and Reynolds number correlation used in the E-NTU method is given in Equations (1) and (2) [7, 8]. Nu v hd (1) 0.6 v / k Re pv

6 The surface-area heat transfer coefficient is denoted by h, where k is the thermal conductivity of air and Re pv is the Reynolds number described by the fluid viscosity, volume equivalent diameter, D v, and the mass flux, G, through the packed bed. The formula for the volume equivalent diameter is given in Equation (4). Re pv GD / (2) v CFD SIMULATION The simulation software used for producing CFD results in this study is the Fluent package of ANSYS. The built in porous media model option is used for the simulation of heat transfer between the air and the rocks and flow of the fluid in the porous media. Heat transfer between air and rocks is difficult to model and to define in the software; therefore assumptions are made to simplify the simulation setup, still imitating reality. Assumptions that are made to simplify the modelling of the porous media geometry are the following: A constant porosity is assigned to the whole geometry. An equivalent particle diameter is calculated based on the average size of the crushed rock. Equivalent particle diameter, D, resembles the rock particles by a sphere with a similar surface area to volume ratio [9]. To measure the area of a non-uniform rock is not a simple and easy task and is very time consuming. An equivalent volumetric diameter can therefore be calculated by only measuring the volume of the particles [10]. D D 6 V / A (3) 6 1 v V pi n i1 p n p 1/ 3 (4) The parameters dictating the fluid flow over the porous media is defined with two main variables, namely the inertial resistance and the viscous resistance terms. Both these terms are described in detail in the User Guide of ANSYS 15.0 [11]. The viscous resistance term is 1/α, and the inertial resistance term is C 2. For the equivalent particle diameter, D p, the calculated volume equivalent diameter, D v, is used to calculate the resistance terms making use of Equations (5) and (6). 2 D p C 2 (6) 3 D p Simulation Setup Simulation of flow over a porous medium involves a lot of turbulent flow over the intricate geometry. The chosen turbulence model used in this simulation is the k-ε turbulence model. More turbulent flows are well described with this turbulent model and it is computationally more realizable [12]. Boundary conditions are chosen to simulate best what would happen in reality. The inlet boundary condition is set to be a velocity inlet. This steady parameter ensures that the inlet mass flux stays constant throughout the charging cycle. A fixed inlet temperature and a fixed velocity maintain the momentum and thermal equilibrium of the energy input to the system while in the charging mode. A pressure outlet is selected to control the fluid flow on the outlet boundary. Ambient pressure surrounding the thermal energy storage is kept constant at a 0 Pa gauge pressure. Walls on the outer circumference of the containment structure and also the center pipe on the inside is assigned a non-slip, stationary wall condition. Flow near the walls of the containment structure will be influenced by the surface friction, but would be insignificantly small as result of the low velocity of the fluid at these regions. The packed bed of rocks is assigned isotropic porous medium properties. The void fraction of 0.45, as shown in Table 2, is assumed to be constant through the whole volume of the computational domain (packed bed). The average void fraction is representative of the void in the packed bed for calculations [2]. (5)

7 A dimensioned drawing of the computational domain can be seen in Fig. 4. Three-dimensional meshing is used to divide the computational domain into 3D cells. A section that illustrates the meshing of the computational domain is shown in Fig. 5. Meshing is finer closer to the center of the computational domain. The cell size is chosen to be an order of magnitude larger than the particle sizes to minimize the computational effort of the simulation. The porous media option in the simulation software uses integrated parameters for the cells that is representative average values for a small volume of particles. This is the general reason for a simplification by using the porous media option [11, 12]. FIGURE 4: Section of the three dimensional computational domain geometry of the porous media FIGURE 5: Two dimensional view of a three dimensional mesh in the computational domain CFD Modelling: Design Considerations Preliminary CFD results are used as a guideline to the experimental layout of the physical structure and also to have a good idea where the region of the hotspot would be situated. Design of the geometry of the containment structure as well as the air ducting can be adapted to accommodate the heat distribution as seen in Fig. 6. The CFD results at this stage are still only qualitative and are not an accurate distribution of the temperature in the packed bed of rocks. The secondary buoyancy effect of the rising hot air can clearly be seen in the plumes moving upward in the center of the packed bed. Low temperature regions in the outer boundary of the packed bed indicate that the storage material is only present for structural purpose. Figure 7 illustrates the path lines of the fluid where the primary and secondary flow patterns can be observed. Two vortices forming in the lower region of the flow is still an undefined flow region. The driver for the flow characteristics should be investigated in future studies. Other layout geometries of a packed rock bed are possible, but a potential low cost option is investigated in this study [8]

8 CFD Results for Temperature Distribution FIGURE 6. Simulated results of the fully charged packed bed directly after charging: Temperatures in Kelvin are just for reference and a qualitative representation of the thermocline in the TES. FIGURE 7: Path lines by colored ID of the flow during the charging cycle. ROCKBED SPECIFICATION Rock type, size and form used in this study have been experimentally tested through thermal cycling to select the most suitable rock for the application of thermal energy storage in previous studies. Thermal capacity of a wide range of different rocks is suitable, but it could not withstand the thermal cycling and subsequently break down after only a few thermal cycles. Properties of a suitable rock type for high temperatures are given in Table 2 [7]. This rock is locally available and the most suitable option for this study. This is the only locally available rock type in the wanted size range. Alternative supply of rocks as a storage material would increase the cost of the material with as result of additional transporting costs

9 TABLE 2. Physical properties of the rocks used in this study Rock properties (Hornfels, Schist) Value Symbol Density 2650 kg/m 3 ρ Thermal conductivity 1.5 W/mK k Heat Capacity 820 J/kgK C p Void fraction 0.45 ε Volume Equivalent diameter 40 mm D v Natural angle of repose 38 θ Average size 53 mm -- FINDINGS The CFD model that is used to produce the results in this paper is still a first investigation of what could happen with the experimental setup when it is in operation. After experimental results have been obtained, the model can be refined to be a more accurate representation of the practical setup. Simulation results showing what the temperature distribution is in the TES material give a good indication of where the critical areas would be to take proper measurements. Design of the containment structure as well as exact placement of the temperature and pressure measuring instrumentation can be deducted from the CFD results. Critical areas that need to be insulated at the base of the TES system can also be identified to install the minimum, but adequate insulation to prevent major losses to the soil underneath. The temperature distribution graphic indicates that the primary drive for the flow in the porous media is the forced convection induced by the fan in the system. A secondary drive for flow direction is the pressure difference induced by the difference in density of the air (fluid). This secondary drive caused by the natural buoyancy can cause instability of the TES system when in the idling (resting) mode. Alternative mitigation strategies could be proposed in future research to deal with the buoyancy and potential loss of energy through air escaping upwards. Future experimental measurements and simulation would focus on the thermocline in the storage media, pressure drop over the porous media and the overall efficiency of the storage system. ACKNOWLEDGEMENTS Thank you to my supervisors, STERG and the NRF for funding. Great thanks to Stellenbosch University for giving me the opportunity to make my contribution to a better future. REFERENCES 1. P. Gauché, Thermal Energy Storage Facility, South African provisional patent 2014/03555 (2014) 2. K.G. Allen, T.W. von Backström, D.G. Kröger, (2015). Rock bed pressure drop and heat transfer: Simple design correlations, ScienceDirect, Solar Energy 115 (2015) AK Steel, Stainless steel data sheet, Accessed April P.J Hughes, The design and predicted performance of Arlington House. MSc thesis, University of Wisconsin Madison, J.A. Duffie, W.A. Beckmann, Solar Engineering of Thermal Processes, Second ed. Wiley, New York, T.E.W. Schumann, Heat transfer: a liquid flowing through a porous prism. J. Franklin Inst. 208, , K.G. Allen, Rock bed thermal storage for concentrating solar power plants, PhD dissertation, University of Stellenbosch, K.G. Allen, T.W. von Backström, E.C. Joubert, P. Gauché, Rock Bed Thermal Storage: Concepts and Costs, SolarPACES Conference October 13-16, Cape Town, South Africa, S. Ergun, Fluid flow through packed columns. Chem. Eng. Prog. 48 (2), (1952). 10. R. Singh, R.P. Saini, J.S. Saini, Nusselt number and friction factor correlations for packed bed solar energy storage system having large sized elements of different shapes. Sol. Energy 80 (7),

10 11. ANSYS 15.0 User guide, July W. Shih, A. Liou, Z. Shabbir, Yang, J. Zhu, A new k-epsilon eddy-viscosity model for high Reynolds number turbulent flows - Model development and validation, Computers and Fluids, 24 (3) (1995), pp