MODELING OPTIMIZATION OF FIXED ARBITRARY CONCENTRATORS Jenni Brito Department of Chemical Engineering Universidad de las Américas-Puebla Santa Catarina Mártir San Andrés Cholula, Puebla, México 72810 Email: jenni@udlap.mx Erick R. Bandala Department of Civil and Environmental Engineering Universidad de las Américas-Puebla Santa Catarina Mártir San Andrés Cholula, Puebla, México 72810 Email: erick.bandala@udlap.mx Brian W. Raichle Department of Technology and Environmental Design Appalachian State University Katherine Harper Hall Boone, NC 28608 Email: raichlebw@appstate.edu ABSTRACT Non-tracking solar concentrating systems may enhance the performance of some solar applications. One application under consideration is solar-induced photocatalytic water purification. To increase purifying capacity it is desirable to have uniformly distributed radiation around the circumference of the proposed cylindrical photocatylist surface. This requirement suggests a reflecting system, and since low manufacturing and maintenance costs are desired, a passive system is sought. Optimization of the reflecting geometry is a challenge. Optical efficiency of 2D arbitrary reflecting surfaces will be modeled using NREL s SolTrace optical modeling software. Modeling parameters will include axis orientation (E/W or N/S), trough parameters, and absorber size and position. The geometry that optimizes concentration and uniformity of illumination will be identified. Raytracing and absorber illumination distribution will be presented. In addition, modeling results and potential implications for water purification systems will be discussed. 1. INTRODUCTION Potential application of solar radiation as driving force for site restoration processes have emerged as very interesting cost-effective alternative in the last decade (1,2). Advanced oxidation processes (AOPs), specifically heterogeneous photocatalysis (HPC), are among the most promising methodologies capable to use solar radiation for water depuration. In the case of solar applications, solar collection is very important if HPC processes are expected to be efficient. The most of solar collecting geometries potentially adaptable for photo-catalytic reactions have already been used successfully for thermal or thermal-catalytic reactions and are well known. However, HPC processes are activated by radiation absorption and the radiation distribution in the photoreactor, where the reaction kinetics must be obtained, is a very important scaling-up parameter that must be known. If light propagation is considered, the geometry of the photoreactor becomes much more important than in thermal reactors. Radiation transport considerations as well as pre-established shapes and sizes of the solar collection geometries place important limitations to the designer choices. Restrictions are even more severe for titanium dioxide because it has strong UV radiation absorption properties and for usual catalyst concentrations, < 10mm of the reaction space turn the reactor totally opaque. 2. REVIEW OF PREVIOUS STUDIES Comparison among different solar collection geometries for HPC processes have been assessed in several previous works (1,3,4). The comparison has been performed between concentrating and non-concentrating solar collecting geometries (i.e. parabolic trough, compound parabolic and tubular ). Some other works have analyzed radiation absorption in the photocatalyst inside the reactor to estimate the relationship between the local volumetric rate of energy 1
absorption (LVREA) and the performance of the photocatalytic process (5). It has been well established that HPC depends most on the photon flux in the receiver and its distribution than in radiation intensity (6). Some authors have suggested that uniform radiation flow density through all the surface of the photoreactor wall is desired in order to have the highest efficiency for the HPC degradation of pollutants in water (7,8). Few works dealing with the radiation flow distribution in the photorreactor (receiver) of solar collector geometries for HPC applications have been published. The aim of this work is to show our results on the modeling for radiation flow in the receiver of three different solar collection geometries (i.e. parabolic trough, V trough and L shape) to determine its uniformity and correlate it with its performance for HPC processes. 3. METHODOLOGY SolTrace optical modeling software (9) was used to assess the ray tracing analysis of the three proposed solar collection geometries (parabolic trough, V trough and L shape collector) to estimate the radiation flow distribution arriving to the photorreactor. The photonic flow was estimated at different incidence angles. In the same way, different focal distances were simulated to determine its effect on the radiation flow in the photorreactor. applications. In this regard, it is interesting to compare the performance of concentrating vs non-concentrating solar collecting systems and correlate it with the solar radiation incidence angle during the day. Figure 1 shows the photonic flow distribution along the receiver distribution in the parabolic trough (PT) solar collector for different incidence angles (0-90º). As shown, the major amount of rays reaching the photorreactor occurs during zenith (0º). In this case two peaks may be easily identified for specific zones of the receiver where the amount of rays impinging may be as high as 570. As expected, important decrease occurs in the radiation flow in the receiver when the incidence angle changes from 0º to any other value and reaches a minimum when the incidence angle is 20º. Further increase in the incidence angle did not produced significant improvement on the ray count reaching the photorreactor. Ray distribution in the PT, however, may not be consider homogeneous for any incidence angle since some portion of the receiver remains with lower radiation flux during any time. In general, despite PT has the capability to produce high amounts of solar radiation into the photorreactor, the desired homogeneous distribution of the photonic flux is not necessarily obtained. 3.1 The optical system SolTrace is a ray tracing model software developed by NREL to model solar power optical systems and analyze their performance. SolTrace is useful to develop new, complex solar optical designs that previously could not be modeled. SolTrace can model parabolic trough concentrators as well as dishes, towers or other unique geometries. In addition, it can model any number of stages containing any number of different elements. It features an extensive variety of available shapes and contours. The software rapidly displays and saves data as scatter plots, flux maps, and performance graphs. 3.2 Data analysis The radiation flow modeled for the absorber in every solar collection geometry assessed in this work was obtained by modeling though Monte Carlo ray tracing 10 6 rays impinging the solar collector at different incidence angles. The results obtained for every case where analyzed as ray histograms in order to allow a better comparative analysis of the results. 4. PRELIMINARY RESULTS As stated earlier, photonic flux arriving to the photorreactor is one of the most important consideration for HPC Fig. 1: Photonic flow in the photorreactor of the PT as function of the incidence angle. The circle in the right top represents the receiver and locates the ray counter zero. Very interesting results were obtained when using the other two non-concentrating solar collection geometries, the V trough (VT) and L shaped (LS) collectors. As shown in Figure 2, despite VT did not achieved high radiation intensity in the photorreactor, the maximum number of rays reaching the receiver was 200, photon flow distribution shows considerably homogeneity when compare with the distribution observed for the PT. It is interesting to note that VT performance remains higher than those observed for PT at high values of incidence angle as it could be supposed considering that PT geometries operation requires solar tracking systems. The general trend in radiation flow distribution in the VT remains quite constant along the 2
receiver surface with low effect of the incidence angle as also shown in Figure 2. receiver but the opposite, VT and LS remains almost constant over the entire range of angles tested. Fig. 2: Photonic flow in the photorreactor of the VT as function of the incidence angle. The overall behavior of the photonic flow distribution in the photorreactor for the LS geometry as function of radiation incidence angle is depicted in Figure 3. In this case, radiation intensity reaching the receiver is quite similar to the values showed for the VT collector with the same important differences with the PT geometry. As discussed earlier for the VT, photon flow along the photorreactor is also quite homogeneous comparable with the distribution observed in Figure 2. Fig. 3: Photonic flow in the photorreactor of the LS as function of the incidence angle. In order to obtain a better comparison of the performance of the three geometries tested along the day, Figure 4 show the amount of radiation missing the photorreactor for every solar collection geometry at the different incidence radiation angles modeled for this work. As previously determined, PT is the geometry with the worst performance when radiation incidence angle changes from 0º to any different value due to the lack of sun tracking. Interestingly, the other two geometries did not present such drastic modification in the value of rays leaving the system without reaching the Fig. 4: Performance comparison among the modeled geometries for the different incidence radiation angles tested. In order to identify the potential repercussions associated with the results obtained from the ray tracing model to estimate the radiation flow distribution on the photorreactor and the performance of the studied optical systems for HPC, data from this work were compared with experimental results reported by Bandala and Estrada (1) for the solar driven HPC process applied to the degradation of two organic compounds, oxalic acid and carbaryl. These authors tested four different solar collection geometries as well as different catalyst loads (TiO 2 slurries) as reported elsewhere (1). Briefly, the PT system consisted of a single Pyrex glass tube, 0.03 m inner diameter, located in the focal axis of a parabolic like collector 0.80 m wide and 0.95 m long for a concentration ratio of 8.5 Suns. The reflector was an aluminum sheet curved to a parabolic shape over a rigid steel structure. The VT consisted of parallel rows of 8 glass tubes 0.03 m inner diameter. Each tube has a back reflector to enhance solar incidence. The reflectors are aluminum sheets shaped as VT with a concentration ratio of 3 Suns. VT are non-concentrating reflectors whose only function is to improve the distribution of solar irradiation around the tube walls. VT collector was 0.1 m wide and 0.95 m long. Solar collection areas were equalized by covering a portion of the parabolic concentrator with a nonreflecting material. Also, equal mass flow rates were used to ensure an adequate comparison. Despite VT does not require sun tracking to operate, the two geometries were mounted together in a two axis, elevation and azimuth, solar tracking system to ensured that the radiation impinging on each collector was the same, in order to obtain the fairest possible comparison. Figure 5 show the results of the heterogeneous photocatalytic process for degradation of oxalic acid using different TiO 2 concentrations. As shown, it is very interesting that the overall performance of the VT in the 3
HPC degradation of oxalic acid is considerably higher than those showed for the PT at high photocatalyst concentrations. It is possible to consider that this behavior is not related with the amount of radiation reaching the photorreactor at the different geometries since, as stated earlier, PT concentration ratio was 8 suns whereas VT concentration rate was 3 suns. Under this consideration, despite direct beam radiation was in the range between 14.3 and 16 W/m 2 and global radiation was between 36 and 42 W/m 2, total UV radiation available at the PT receiver is higher than those arriving to the VT receiver. intensity and TiO 2 loads shadowing effect have been also observed mainly for non-homogeneous photonic flow distributions. This shadow effect will generate that the photocatalyst particles inside the photorreactor may not being effective for the HPC process because lack of radiation absorption due to other particles absorbing, reflecting or scattering most of the radiation reaching the edge of the photorreactor. Fig. 5: Degradation profiles for oxalic acid under solar driven HPC as function of photocatalyst load using the PT and VT geometries (adapted from (1). It is important to note that at low TiO 2 loads the performance of both geometries is quite similar and differences are significant only when photocatalyst concentration is higher than 0.05 g/l. This is possibly an effect related with the photonic flow distribution. Oxalic acid is an organic compound relatively easy to oxidize because its only oxidation product is carbon dioxide, a very stable molecule, generating that the reaction is thermodynamically allowed. At no or low TiO 2 concentration, HPC process may be considered negligible since few radiation absorption occurs. Under these conditions, photolysis of oxalic acid should be important mainly in the PT collector where an important amount of UV radiation is available. As the catalyst concentration rises, the HPC process becomes more important for the VT whereas in the PT may not increase accordingly due to the poor radiation distribution in the photorreactor. As it has been reported previously (7,8), HPC reaction rate may increase linearly with radiation intensity for low solar concentration rates and with the square root of the radiation intensity for medium solar concentration rates. When photocatalyst concentration increases, besides the effect of radiation intensity, distribution of photonic flow becomes also important. It has been suggested that, as the distribution of photonic flow goes far from uniform, performance efficiency of the solar collector will be lower and the kinetic constant of the degradation reaction will decrease (7) as observed in Figure 5. Additionally, at high radiation Fig. 6: Degradation profiles for carbaryl under solar driven HPC as function of photocatalyst loadusing the PT and VT geometries (adapted from (1)). All the phenomena described for oxalic acid becomes more evident when using another higher refractory organic compound. Figure 6 show the degradation profile of the pesticide carbaryl when submitted to solar driven HPC under the same experimental conditions described previously for oxalic acid and using the same solar collection geometries. As shown, since carbaryl is a chemical quite hard to degrade, differences between the two geometries tested is noticeable even at low photocatalyst concentrations. In this case, as observed for oxalic acid, the same shadow effect is noticeable at high TiO 2 concentrations related with the poor homogeneity in the PT geometry compared with VT. Despite no test were performed for the LS optical system, it is reasonable expecting this geometry may behave similar to the trend showed by the VT anytime photonic flow distribution and radiation intensity in both geometries was found alike using the SolTrace model. In this case, however, the analysis must be carefully carried out because to the opposite with PT and VT, LS does not possess symmetry and its application would be restricted by the radiation incidence angle. 4
photonic flow distribution seems to be a very important factor on the photorreactor efficiency when HPC process are tested. When the differences in the photonic flow distribution where compared with previously reported experimental results for similar geometries used for solar driven HPC processes, the correlation among this factor and the performance of the optic system was evident considering both, organic compound degradation and kinetic values. The results of this work encourage us to continue with this research in order to achieve novel ensigns on the estimation of scaling-up parameters for the improvement of solar driven HPC and its application to site restoration. Fig. 7: Kinetic values for oxalic acid solar driven HPC degradation as function of photocatalyst load using the PT and VT geometries (adapted from (4)). When HPC degradation reaction kinetics is taken account, the differences between the two tested geometries are even most evident. In a previous work, Bandala et al., (1) compared different optical geometries for oxalic acid solar driven HPC degradation. In their work these authors proposed a kinetic model for the chemical HPC degradation as function of radiation absorption by the photocatalyst particles in the reaction mixture and compared the model and experimental results for four different solar collection geometries. Figure 7 show the rate constant estimated for the HPC degradation of oxalic acid as function of catalyst concentration. It is interesting to note that, when radiation absorption is considered in the kinetic process, the gap between the two optic systems increases. For this case, k values obtained for the VT are higher than those determined for the PT even for low photocatalyst loads and maintains as the catalyst concentration increases up to the maximum catalyst load tested. In this case, we think that photonic flow distribution is also a very important consideration as we have previously stated and the performance of the solar collectors is importantly affected by all the parameters previously analyzed. 5. SUMMARY AND CONCLUSIONS Ray trace modeling was successfully carried out for three different solar collection geometries for application in heterogeneous photocatalytic processes. Results obtained from the modeling suggested that photonic flow distribution in the receiver of the geometries tested vary depending on the shape of the reflective, the radiation incidence angle and the surface of the photorreactor. Parabolic trough was identified as the one with the highest radiation intensity with incidence angle equal to zero but with the most variable photonic flow distribution. V trough and L shaped geometries where remarkably lower in radiation intensity but more homogeneous regarding to the radiation flow distribution along the receiver. As previously proposed, 6. REFERENCES (1) Bandala, E.R. and Estrada C.A. (2007). Comparison of solar collection geometries for application to photocatalytic degradation of organic pollutants, Journal of Solar Energy Engineering, 129, 22-26. (2) Quiroz, M.A., Bandala E.R., and Martínez-Huitle C. (2011). Advanced oxidation processes (AOPs) for removal of pesticides from aqueous media. In: M. Stoycheva (Ed.) Pesticides- Formulations, effects, fate. In-Tech Press. (3) Cassano, E.C., and Alfano O.M. (2000). Reaction engineering of suspended solid heterogeneous photocatalytic reactors, Catalysis Today, 58, 167-197. (4) Bandala, E.R., Arancibia, C.A., Orozco, S.L. and Estrada, C.A. (2004). Solar photoreactors comparison based on oxalic acid photocatalytic degradation, Solar Energy 77(5), 503-512. (5) Arancibia, C., Bandala, E.R. and Estrada, C.A. (2002) Radiation absorption and rate constants for carbaryl photocatalytic degradation in a solar collector, Catalysis Today 76 (2-4), 149-159. (6) Blanco, J., Malato, S., Estrada, C., Bandala, E.R., Gelover, S. and Leal, T. (2004) Water detoxification by heterogeneous photocatalysis: State-of-the-art. In Blesa, M.A., Sánchez, B. (Eds.) Pollutants Elimination by Heterogeneous Photocatalysis. Editorial CIEMAT. Madrid, Spain. (7) Curco, D., Malato S., Blanco J., Gimenez J. and Marco P. (1996) Photocatalytic degradation of phenol: Comparison between pilot-plant-scale and laboratory results, Solar Energy, 56, 387-400. (8) Gimenez, J., Curco, D. and Queral, M.A. (1999) Photocatalytic treatment of phenol and 2,4- dicholotphenol in a solar plant in the way to scaling-up, Catalysis Today, 54, 229-243, (9) Wendelin, T., Lewandowski, A. and Dobos, A. (2011). SolTrace, Version 2011.7.5. Developed by the National Renewable Energy. 5