Abstract
This research details the design, fabrication, and partial testing of a concentrated solar receiver and an air-cooled heat exchanger. The solar receiver and heat exchanger have been fabricated for use in an experimental system that uses the supercritical carbon dioxide Brayton cycle. They are coupled with a Science Applications International Corporation (SAIC) solar dish 250× concentrator located on the University of Nevada, Las Vegas campus. The purpose of this solar-powered supercritical CO2 system is to function as a testbed for testing the cycle, system components, and alternate system configurations. Photographic flux mapping of the dish showed peak solar flux just above 200× and is used to appropriately size the receiver. Sun tests of the tubing, receiver, and air-cooled heat exchanger were performed achieving fluid temperatures in the range of 973 K (700 °C) using nitrogen in an open loop at low mass flowrates, and above 1173-K (900 °C) receiver wall temperatures in a no-flow case.
Introduction
As worldwide energy consumption trends upward, developing more efficient, clean energy systems with decreased environmental and geographical footprints is necessary in energy generation research and development. In an effort to accomplish these goals, power generation systems other than the traditionally used Rankine cycle are considered. The idea of using supercritical power cycles has been discussed since before Feher [1] published his theoretical conclusions in 1968. Advantages of using a supercritical carbon dioxide Brayton cycle, or SCO2 Brayton cycle, include a vast decrease in component size, an increase in thermal efficiency, blade corrosion resistance, and the ability to eliminate nearly all system water consumption when coupled with dry cooling. Additionally, carbon dioxide is preferable due to its abundance, noncorrosiveness, incombustibility, and low critical temperature and pressure when compared to some other gases and working fluids [1,2].
SCO2 cycle performance can be attributed to two factors: temperature and fluid properties. The performance is calculated using the temperature difference between the hot flow entering the turbine and the cold flow entering the compressor, and the specific heat at each state. The thermophysical properties of carbon dioxide in the supercritical phase, particularly near but above the critical point, vary significantly with small changes in temperature. Concentrated solar is capable of achieving temperatures in the range of 773–1073 K (500–800 °C) [2,3]. With the use of recuperators the expected receiver inlet temperature range is 573–723 K (300–450 °C). With cold-side temperatures above critical in the range of 304–318 K (31–45 °C), the temperature difference resides in the 741–1041 K (468–768 °C) range [4]. A high concentrated solar SCO2 Brayton cycle system, shown in Fig. 1, is developed at the University of Nevada, Las Vegas (UNLV). It uses the 100 kW Science Applications International Corporation (SAIC) Solar Dish Concentrator with a nominal concentration ratio of 250× located on the UNLV Center for Energy Research (CER) site. This SCO2 Brayton cycle system is being built for testing the SCO2 Brayton cycle, different system components, and different system configurations. Major components include two purchased printed circuit heat exchangers which serve as recuperators, a UNLV-built turbine-compressor, a solar receiver, heat rejection coil, and a purchased booster pump. The system also includes a Programmable Logic Controller package that handles all temperature, pressure, mass flow, turbine speed measurements, and system operations.
Design
The solar receiver is designed to fit within the receiver package dimensions of the SAIC dish concentrator. The receiver area measures 1 m2. The material used as the flux-mapping collection surface and receiver insulation is made of two acid-resistant ultra-high-temperature flexible silica insulation sheets that are supported on the back such that they are flat and held in place in all orientations. Photographic flux mapping [5] is used for determining the flux distribution. This method uses digital images collected from the SAIC dish concentrator operating on-sun. The UNLV system operating on-sun is displayed in Fig. 2.
Digital Image Collection.
The Ho and Khalsa’s method for photographic flux mapping [5] requires high-resolution digital images of the sun and the receiver area, the direct normal irradiance at the time of image collection, and the reflectivity of the receiver. Gray-scale images of the sun and the concentration area were collected during on-sun operation using a 14.2-megapixel, Nikon D3100 camera, having a 55-mm focal length, with neutral-density filters. Neutral-density filters are used to decrease the amount of light entering the camera without altering the image. The image of the sun required four Hoya ND8 and one ND4 filters, providing optical densities of 0.9 and 0.6, respectively. To collect the receiver image, the filter combination was reduced to two ND8 filters, providing an optical density of 0.9 each. The total optical density is given by adding the optical densities from each filter used for each image together. The expected total optical density is 4.2 for the sun image and 1.8 for the receiver flux-mapping surface image. The optical density of each filter is measured with a spectrometer by taking a reading of the sun with and without each filter. The spectrometer determined the total optical density to be 4.59 for the filter combination used to collect the sun image and 1.98 for the combination used to collect the receiver image. The resulting sun image and receiver area flux-mapping image, using the described filters, are displayed in Figs. 3 and 4, respectively.
Image Conversion.
The script converts the imported sun and flux-mapping receiver images into 16-bit gray-scale images (Figs. 3 and 4). Each pixel has a number from zero to 65,535, where black is zero and white is 65,535. The derivative of each pixel in the sun image is calculated to isolate those pixels making up the sun. Sun image pixel normalization is completed in the script, filtering pixels with low-value derivatives to determine the edge of the sun within the image. The result is used to determine the sun pixel radius, center point, and area. Each pixel value is then converted into a relative irradiance value that yields a pixel summation equal to the direct normal irradiance collected by a pyrheliometer at the time of image collection. Figure 5 is the resulting pixel conversion of the sun image, with each pixel displaying a color-coded value corresponding to a portion of the total direct normal irradiance at image collection, 990 W/m2.
The angle subtended by the sun, γs, is found by using the distance from the sun at the time of image collection. This information is provided by the NASA Jet Propulsion Laboratory Horizons System (JPL Horizons) [6]. JPL horizons provides a high-precision Earth Orientation Parameters model. The observation location for the model is entered as the location of the UNLV CER, 115°08ʹ39.6ʺW and 36°06ʹ51.1ʺN at 617 m (2025 ft). Target angular diameter output, in arcseconds, is set to use the date and time corresponding to the minute the sun image was collected. The radius of the sun is estimated to be 696,000 km. The angle between the camera and flux-mapping receiver normal is calculated by trigonometry and physical measurements.
All of these parameters are entered into the flux-mapping matlab script and a resulting image is generated which is color-coded to differentiate the levels of concentrated flux on the flux-mapping receiver (see Fig. 6). The error sources and their estimated ranges were described by Ho and Kalsa [5]. Each source of error was determined through either the manufacturer's stated uncertainty or those mentioned by Ho and Kalsa [5]. The uncertainty of the camera was estimated to be ±3%, the normal incidence pyrheliometer ±1.4%, neutral-density filters ±5%, insulation reflectivity ±5%, and the sun shape ±4%. The square root of the sum of the squares of each uncertainty gave a total uncertainty for the flux measurement of ±8.77%.
Receiver-Size Design.
The solar flux distribution shown in Fig. 6 provides both the maximum concentration ratio achieved on a clear day and what the minimum size of the solar receiver must be to avoid spillage loss. The beam is nearly centered within the solar concentration area in a rounded square pattern, achieving a solar flux of over 2 × 105 W/m2 ± 1.75 × 104 W/m2 from the center to approximately 0.25 m (10 in.) vertically and 0.20 m (8 in.) horizontally. The intensity decreases concentrically until reaching one sun between approximately 0.33 and 0.41 m (14 and 16 in.) from the center point. The line shown horizontally across the center of the collection area is the seam between the two pieces of high-temperature insulation material that covers the collector area.
To minimize flux spillage, it is desired to make the receiver as large as the concentrated beam. A slightly off-center cross-sectional flux profile is used to avoid the seam in the receiver material. The flux profiles are displayed for the horizontal cross section in Fig. 7 and for the vertical cross section in Fig. 8. Zero is the center point of the receiver area, corresponding to the central pixel location. Horizontally and vertically, the data show that the beam is not perfectly centered. The skew is due to errors in solar tracking. The SAIC dish concentrator tracks the sun using ephemeris equations to calculate the sun's position and two encoders to match the tracker position in azimuth and elevation. The tracker does not have any other sensors to provide feedback to correct any tracking errors.
Variables to consider for the design of the solar receiver are power intercept, pressure drop, existing receiver package dimensions, and ease of manufacturing. The receiver area flux distribution shows a rounded square beam; therefore, this geometry is considered. Fabricating concentric circles is easier than that for concentric squares; therefore, circular geometry is also considered. To establish power loss from beam spillage as a function of geometry and size, the SAIC final report [7] which measured power interception as a function of a square receiver size is used. These data are compared with varying radii and square-side dimensions of receivers with round and square design geometries for the most recent flux image measurement. The data comparison is analyzed up to a 0.46-m (18 in.) half-length. The resulting comparison, described in Fig. 9, shows that a square receiver with a 0.38-m (15-in.) square-side half-length, or that is 0.76-m (30 in.) square, would result in acceptable power loss compared with the expected power intercept stated in the SAIC report [7]. Power loss is greater with a circular design of equivalent half-length due to geometric corner loss. At 0.38 m (15 in.), the power intercept loss for the circular design has a 16.7% ±2.5% power loss. At the half-length estimation of 0.41 m (16 in.), the circular design power loss expectation decreases to 9.8% ±2.7%. The latest measurements show the receiver size needs to be larger than what was found in the SAIC report [7]. This is probably due to mirror degradation/cleanliness and small changes in dish structural alignment over time.
Fabrication
Based upon the data collected, existing receiver package dimensions, and ease of fabrication, a circular geometry with a 16-in. radius is selected. To reduce beam reflection loss and to allow for space for fittings between sections, a change in height was given to the receiver as it spirals inward. Thus, the final solar receiver design is a bugle shaped spiral with an overall diameter of 0.81 m (32 in.). To reduce thermal losses, the gas flow inlet of the receiver will start on the outer edge where the flux is lowest and exits in the center where the highest concentrated flux is located.
Flux results indicate receiver material temperatures may reach up to 1273 K (1000 °C) or more; thus, tube material is chosen accordingly. While Inconel provides extreme temperature tolerance it is very difficult to form and it is desired to be able to purchase off the shelf components to assemble the receiver. The receiver will also only be for experimental use and will not be used for daily operation. For these reasons, thick wall stainless steel (SS) 316 is chosen. Swagelok SS 316 tube with an outer diameter of 12.7 mm (0.5 in.) and a wall thickness of 2.11 mm (0.083 in.) meets the temperature requirements and provides a standard pressure rating of 46.2 MPa (6700 psig). Accounting for 1273-K (1000 °C) temperature degradation, 76% of the standard value, the tube maintains a pressure rating of 35.1 MPa (5092 psig).
A custom bending tool is fabricated to build the receiver. The custom bender is capable of bending the tube in spirals with adjustable radii. A photograph of component fabrication using the UNLV Custom Radii Tube Bender tool is shown in Fig. 10. This tool is used to create the shape of the receiver, which is then secured using SS wire, inset in a cylindrical housing of sheet metal, and wrapped in insulation, producing the final product shown in Fig. 11.
Figure 12 is a backside view of the solar receiver. In this view, the expansion loops are visible. These expansion loops serve three purposes: they are a method of connecting tube lengths and also allow for the expansion of the material under extreme temperatures, and provide a shielded location for the tube fittings. The tubes shown in a perpendicular orientation to the spiral tubes provide structural support and do not house flow.
The fabricated solar receiver is 0.83 m (32.5 in.) in diameter. The calculation of the concentrated solar flux intercepted by the UNLV solar receiver is 91.6%. If the receiver is assumed to have developed a thermal oxide layer, it would have an absorptivity of 0.9 [8]. Therefore, if losses are ignored, the maximum percentage of the concentrated solar flux captured by the UNLV solar receiver is 82%.
Experimental Test Components
Experimental testing of the solar receiver is coupled with a custom heat rejection heat exchanger. The air-cooled heat rejection heat exchanger (ACHEX) is designed to remove all of the heat accepted by the solar receiver. Its goal is to reach a gas temperature as close to ambient as possible. Sizing of the ACHEX is estimated by an Engineering Equation Solver (EES) model. Once the size is determined, it is decided that a circular spiral shape will provide a condensed design with decreased fluid friction losses compared to other geometries. Additionally, this circular spiral shape will allow additional rings to be added if needed. The custom bending tool fabricated for the solar receiver is used to construct the ACHEX, displayed in Fig. 13. Connections made between tube lengths are combined with expansion loops for flexibility.
In normal operation the UNLV SCO2 Brayton cycle system will use carbon dioxide in the supercritical phase as the working fluid; however, for open-loop testing and in all leak testing, nitrogen was used as the test fluid. The initial receiver tests using nitrogen will not match the pressures and flowrates that the final system would achieve. The pressure at the receiver would be about half of what it is expected to be and the mass flowrate much lower at 5% of the maximum. While these test conditions were not ideal, it allowed for the testing of several components at once such as the temperature sensors, receiver tube tie-downs, receiver, and the ACHEX (at least at low flowrates). In order to test the system at design conditions, it would require the turbo compressor, booster pump, automated valves, and the programmable logic controller (PLC) to be installed, all of which were not available at the time. These initial tests at the very least indicate whether any design changes would need to be made with the receiver, ACHEX, or any of the temperature sensors.
Experimental Testing and Analysis
Solar Receiver Material and Temperature Test.
The solar receiver was tested in two phases. The first was with a small test segment of receiver tubing in a three-coil pass configuration for material temperature tolerance in the no-flow condition. The receiver test section, shown in Fig. 14, was mounted onto the solar receiver area. One thermocouple was placed on the outer wall (T1), and two more were inserted into the tube of the small receiver test segment (T2 and T3). The SAIC dish was then positioned on-sun. The results, displayed in Fig. 15, show that all three thermocouples reach 1123 K (900 °C ±8.4 °C) in less than 5 min. Error bars are shown for the T2 thermocouple but are difficult to see because of the large scale. The direct normal irradiance during the test is approximately 930 W/m2 ± 13 W/m2.
The thick-walled, stainless steel (SS316) tube is capable of withstanding the high solar flux given by the SAIC dish concentrator for short time periods. The test segment displayed discoloration; however, the segment did not display visible deflection or evidence of material integrity loss. Second, this test shows that the system can reach the desired temperatures. It also demonstrated the ability of the stainless steel tie wires used to secure the tubes and the reliability of the stainless steel jacketed thermocouples installed around the fabricated receiver. Again, this receiver is not designed to be used for daily operation but for use in a testbed system for SCO2 cycle experiments.
Solar Receiver Heat Addition and Heat Rejection Test.
The second phase of solar receiver testing occurred in conjunction with a heat rejection test of the ACHEX. In this test, an open-loop configuration with nitrogen gas was used. The goal was to gather initial performance data from the solar receiver and to determine whether the designed size of the ACHEX sufficiently rejects heat. RTDs and pressure transducers were used at the receiver inlet, receiver outlet, which feeds the ACHEX, and ACHEX outlet. A Kulite high-temperature pressure transducer is used at the receiver outlet location. Eight thermocouples were also placed on the backside of the receiver at locations shown in Fig. 16.
Concentrated flux began hitting the receiver at 10:09 and was removed from the receiver after 10:16 a.m. while nitrogen gas was continuously flowed from the tank and through the receiver and ACHEX to the outlet. The flow was manually controlled with valves to try to maintain a target pressure of 6.9 MPa (1000 psig). The mass flowrate and the three RTD data results are displayed in Fig. 17. The flowrate, measured by a Coriolis flowmeter, is shown on the right vertical axis. Error bars are shown for the RTDs and mass flowrate but are difficult to see due to scale. The nitrogen gas passing through the UNLV receiver reaches a temperature of 702 °C ± 3.5 °C. The RTD on the receiver outlet was damaged during this test by exceeding its temperature rating. This RTD was later replaced by a thermocouple with a much higher temperature rating. The receiver inlet and the ACHEX outlet temperatures maintain a similar temperature throughout the test. This shows that the ACHEX is capable of sufficiently cooling the solar-heated, high-temperature gas at low flowrates. Further, it is shown that once the system is in the on-sun position, a high receiver gas temperature can be achieved in less than 7 mins.
Figure 18 shows the back of receiver temperatures at eight locations. Discrepancies between thermocouple readings on a single tube are most likely due to thermocouple surface contact and flux distribution differences. All critical gas temperatures and pressures are measured in the flow stream. Also apparent is that the receiver had not yet reached a steady-state temperature before flux was removed and that the receiver has a significant thermal mass which may aid in smoothing out intermittent clouds.
Discussion and Conclusions
After on-sun testing, the effect of the high concentrated flux can be seen on the receiver surface. The pattern reflects the results visible in the photographic flux-mapping data. This is evident in a visual comparison of Fig. 11 with Fig. 6. The lack of significant color change in the outer two rings of the solar receiver demonstrates that the receiver was appropriately sized. The ACHEX also appears to be sufficient in rejecting heat to the ambient. Additionally, the receiver heat capacity shows promising results for smoothing out operations during transient clouds. Continuation of this research will provide full system data and analyses.
Acknowledgment
Contributions to this work were also made by Terry Kell, Mechanical Engineering Machine Shop Director, University of Nevada, Las Vegas.
Funding Data
National Science Foundation (Grant No. IIA-1301726; Funder ID: 10.13039/100000001).
Nomenclature
- i =
subscript describing a specific pixel
- fR =
receiver image attenuation factor
- fS =
sun image attenuation factor
- rsp =
sun image radius in number of pixels
- ED =
direct normal irradiance (W/m2)
- ER =
irradiance on a receiver element (W/m2)
- VC =
pixel value as voltage per square pixel unit
- ρR =
receiver reflectivity
- σ =
standard deviation
- γS =
angle subtended by the sun (radians)