Optimizing solar disinfection: Cost-effective solutions with simple collectors

Optimizing solar disinfection: Cost-effective solutions with simple collectors M. Martín-S´ omer a,* , J. Moreira a , J. Moreno a , J. Marug´ an a,b a Chemical and Environmental Engineering Group, Rey Juan Carlos University, C/ Tulip´ an s/n, M´ ostoles, Madrid 28933, Spain b Instituto de Investigaci´ on de Tecnologías para la Sostenibilidad, Rey Juan Carlos University, C/ Tulip´ an s/n, M´ ostoles, Madrid 28933, Spain ARTICLE INFO Keywords: SODIS Water treatment Ray Tracing Solar Collectors Solar Radiation ABSTRACT Access to safe drinking water remains a critical global challenge, especially in low-income regions. Solar water disinfection (SODIS) offers a sustainable and low-cost treatment alternative, though its implementation is often limited by long exposure times and low treatment capacity. This study investigates the use of simple flat collectors (featuring one-, two-, or three-sided flat reflectors) as a cost-effective means to enhance SODIS performance, as compared with the widely used compound parabolic concentrators (CPC). Over 1300 collector configurations were evaluated using a ray tracing simulation tool, considering solar incidence, latitude (0◦–50◦), and varying cloud cover. The collectors were assessed under two criteria: yearly solar dose and yearly solar dose per unit of projected area. Results show that, under the first criterion, one-sided and two-sided flat collectors can surpass the performance of a standard CPC at higher latitudes, despite being much cheaper and easier to fabricate. Under the second criterion, all simple collectors outperform the CPC, highlighting their superior efficiency in terms of space usage. Experimental validation through actinometry and E. coli inactivation confirmed the simulated predictions. However, achieving such performance requires precise geometric design tailored to the target location. This study demonstrates the potential of well-designed, simple collectors to provide an affordable and efficient alternative for decentralized water disinfection. 1. Introduction On September 25, 2015, 193 countries committed to achieving the 17 Sustainable Development Goals promoted by the United Nations in the 2030 Agenda. Among these objectives, the sixth one clearly focuses on the availability and sustainable management of water, as well as ensuring adequate sanitation worldwide [1]. However, the global water crisis affects communities unequally and is especially severe in those with fewer economic resources [2]. Data provided by UNICEF and the World Health Organization are clear: one in three people worldwide lacks access to clean water, resulting in 829,000 deaths per year that could be prevented with adequate water supply, sanitation, and hygiene measures. Within this tragic scenario, it is particularly alarming that 297,000 of these deaths are children under five years old, meaning that more than 800 children die every day from diarrheal diseases associated with poor hygiene [3]. In this context, the urgent need to address the problem of the availability of clean water and the lack of adequate sanitation in many parts of the world becomes evident. Poor water quality can lead to the transmission of a multitude of diseases, including cholera, dysentery, hepatitis A, typhoid fever, and poliomyelitis [4–6]. Therefore, it is of vital importance to research and develop methods for disinfecting contaminated water using the resources available in each geographical location. Among the initiatives to address this issue is the Solar Disinfection method (SODIS), which has emerged as a sustainable and low-cost alternative for eliminating pathogens present in water, including bacteria [7], viruses [8] and protozoa [9]. This method utilizes solar radiation to disinfect water by using transparent bottles exposed to the sun for a specific duration [10]. However, there are certain limitations that encourage improvement and development. Some disadvantages include the long exposure times required (6–12 h), the opacity of some materials that attenuate the UVB radiation, causing direct mutations in the DNA molecule, and the low treatment capacity (usually utilizing bottles with a 2-L capacity) [11, 12]. To overcome this latter limitation, previous studies have investigated the possibility of working with continuous flow reactors at low Abbreviations: CFU, Colony Forming Unit; CPC, Compound Parabolic Concentrator; PTC, Parabolic Trough Collector; SODIS, Solar Water Disinfection. * Corresponding author. E-mail address: [email protected] (M. Martín-S´ omer). Contents lists available at ScienceDirect Journal of Environmental Chemical Engineering journal homepage: www.elsevier.com/locate/jece https://doi.org/10.1016/j.jece.2025.120718 Received 20 October 2025; Received in revised form 12 December 2025; Accepted 13 December 2025 Journal of Environmental Chemical Engineering 14 (2026) 120718 Available online 17 December 2025 2213-3437/© 2025 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/bync-nd/4.0/ ). flow rates or with large-volume PET containers (20–25 L) [13–15]. The results showed that biocidal action decreases in both cases. In the first case, this is due to the shorter exposure times, while for the containers, it is largely attributed to the heterogeneity of irradiation in different parts of the container, which decreases as the residence times become longer [14]. Most of these limitations can be mitigated by increasing the amount of UV radiation that reaches the water to be treated. For this purpose, reflective surfaces capable of concentrating solar radiation can be incorporated [16]. Among the most common concentration systems, the parabolic trough collector (PTC) and the compound parabolic collector (CPC) stand out. The PTC consists of a collector whose reflector follows the shape of a parabola along a cylindrical channel. The radiation is reflected and concentrated towards the receiving tube (through which the water circulates) located on the focal line of the parabola. This type of collector only allows direct solar radiation to be concentrated, so its use is limited to conditions of low diffuse radiation. To maximize the concentrated radiation, it is possible to connect them to a solar tracking system that maximizes the direct solar radiation received, although this would entail a significant increase in costs [17]. On the other hand, the CPC is a static collector whose reflecting surface is arranged in such a way that the normal at each point of the collector is tangent to the circumference that constitutes the section of the tube. The main characteristic of these collectors is that they can concentrate both direct and diffuse radiation, eliminating the need for a solar tracking system [18]. In both cases, the use of this type of system enables the capture of more incident radiation for the process, thereby increasing its efficiency. However, this improvement also represents a significant increase in the total cost of the process, as the collectors are typically constructed with aluminum, and manufacturing to obtain the necessary parabolic shape for achieving the correct reflection of sunlight is not simple [19]. Consequently, the increase in the cost of water treatment may render it an unaffordable expense for low-income communities. This study suggests eliminating the machining process through the use of collectors consisting of flat reflectors. The use of flat reflectors can lead to a multitude of simple collectors with various configurations, degrees of complexity, or capacity. In this study, the need to reduce costs is recognized, and various geometries that can serve as alternatives to current standard collectors are thoroughly investigated by evaluating the effectiveness of each configuration. The concentration effectiveness of one-sided, two-sided, and three-sided flat collectors was studied based on their geographical location and solar position using a ray tracing tool developed previously [20,21]. Additionally, optimal collectors of each type for the location of Universidad Rey Juan Carlos facilities were constructed. Experimental tests were conducted, including actinometric reactions to measure the amount of concentrated light and microorganism inactivation assays using Escherichia coli as a representative fecal indicator organism, with the aim of validating the delivered UV dose in the different geometries. 2. Materials and methods 2.1. Reactor configuration Within each type of flat collector (one-, twoand three-sided), there are a multitude of configurations depending on parameters such as collector width (w), the angle between the flat reflectors (β), or the reflector depth (d) (Fig. 1). The study of the efficiency of thousands of different cases was carried out by varying these parameters. For the onesided flat collector, f (distance from the center of the receiver tube to the collector) values from 1/2⋅D to 3/2⋅D were studied (1/10⋅D step). On the other hand, various collector widths (w) to receiving tube diameter (D) ratios ranging from 1 to 3 were explored (1/5⋅D step). In the case of the two-sided collector, the study included w/D ratios from 1 to 3 (in 1/5⋅D steps), f values ranging from 7/10⋅D to 3/2⋅D (in 1/ 5⋅D steps) and d values ranging from 1/2⋅D to 5/2 D (in 1/5⋅D steps). Within this studied range, cases that caused collector-reactor intersection were discarded, considering the expression: β=tan−1W/2 h>sin−1D/2 f Nomenclature Latin Letters bBottom side of the 3-sided collector (m) cSpeed of light (m⋅s⁻¹) C Fe³ ⁺ Concentration of Fe³ ⁺ in the actinometric solution (mol⋅m⁻³) C Fe 2 + Concentration of Fe 2 ⁺ produced in the actinometric reaction (mol⋅m⁻³) dCollector depth (m) DDiameter of the reactor tube (m) D act Radiative dose absorbed by the reactor tube determined by actinometry (J⋅m −2 ) D rad Radiative dose in the normal direction to the collector surface (J⋅m⁻²) D tube Radiative dose in the reactor tube derived from radiometer readings, DCF values and optical properties (J⋅m⁻²) D UV UV dose received in the reactor tube throughout the experiments (J⋅m⁻²) fDistance from the centre of the receiver tube to the collector (m) GHI λ Spectral global horizontal irradiation (W⋅m⁻ 2 ⋅nm −1 ) GR Geometric concentration ratio (-) HPlanck’s constant (J⋅s) NNumber of sides of simple collectors N Av Avogadro’s number (mol⁻¹) q rad Flux measured by radiometry on the opening surface of the collector (W⋅m⁻²) r λ Spectral reflectivity of the collector material (-) tReaction time (s) TIR Total incident radiation (W⋅m⁻²) T λ Spectral transmittance of the reactor tube (-) S tube External surface of the reactor tube (m 2 ) VF Validation factor (-) V sol Volume of actinometric solution used in the actinometrical reaction (m³) V tube Volume of the reactor tube (m³) wCollector width (m) ZCorrection factor accounting for the spectral range of the UV radiometer (290–390 nm) with respect to the total solar spectrum (-) Greek Letters α Solar incidence angle (º) βAngle between collector sides (º) κ λ Calculated spectral radiation absorption coefficient (m⁻¹) κ λ * Specific spectral solar absorption coefficient (m 2⋅ mol⁻¹) κ solar * Specific solar absorption coefficient (m 2⋅ mol⁻¹) λ solar Representative solar wavelength (nm) ΦSolar quantum efficiency (-) Φ λ Spectral quantum efficiency (-) φTopocentric azimuth angle (eastward from north) (rad) M. Martín-S´ omer et al. Journal of Environmental Chemical Engineering 14 (2026) 120718 2 For the three-sided collector, the study covered w/D ratios from 1/ 2–3 (in 1/2⋅D steps), f values ranging from 3/10⋅D to 3/2⋅D (in 3/10⋅D steps), d values ranging from 1/2⋅D to 3/2D (1/5⋅D steps) and b values from 1 to 3⋅D (2/5⋅D steps). Again, the collectors that caused the collector-reactor intersection were discarded, considering the expression: β=tan−1w/2−b/2 h>sin−1D/2−b/2 f The total number of reactor configurations studied were 121, 605 and 1080 for the one-, twoand three-sided collectors, respectively, which was reduced to 121, 269 and 764, respectively, after discarding the non-viable configurations. 2.2. Ray tracing simulations A Microsoft Excel® Ray Tracing tool was developed in previous work [20], which allows the calculation of the Dynamic Concentration Factor (DCF) of any 2D geometry of collectors. To facilitate a rapid evaluation of collectors across a wide range of geometric parameters, a C+ + open-source ray tracing tool is presented as part of this work [22]. The use of a lower-level, compiled programming language increased the computational speed by almost four orders of magnitude (approximately 8000 times faster), reducing significantly the computational time required for the evaluation of a high number of reactor configurations. For all collectors, the angle of solar incidence was varied between 1 and 90º, the optical density was set to 0 (as in the SODIS process for clear water) and a reflectivity of 0.9 was set corresponding to anodized aluminum material. Different collector configurations were studied as shown in Table 1. The total number of collector configurations initially generated was 121, 605, and 1080 for the one-, two-, and three-sided collectors, respectively; after discarding the non-viable configurations that produced collector–reactor intersection, 121, 269, and 764 configurations, respectively, were retained and used in the simulations. Once the corresponding concentration factors were determined for all the proposed reactors and considering the position and solar radiation for each day of the year and location [23], the yearly concentration efficiency of each reactor was calculated for latitudes ranging from 0º to 50º. To ensure that the results accurately represent the conditions of different locations, varying degrees of cloud cover were considered, ranging from 0 to 0.75 in increments of 0.25, with the corresponding dispersion and attenuation values for each level of cloud cover [23]. The efficiency was evaluated based on different criteria, including solar dose and dose per unit area. In this sense, the productivity of each reactor is represented by the yearly UV dose obtained for each latitude and cloud-cover scenario, which allows a direct comparison between geometries and can be used as a basis for sizing installations according to the radiation requirements of a given application. For each of these generic latitudes, the Maximum Theoretical Irradiation (MTI) was used, which is independent of longitude and the meteorological conditions of specific locations. The results obtained were compared with the concentration values that would be obtained with a standard CPC reactor with a concentration factor of 1. 2.3. Reactor manufacturing and experimental setup To validate the simulation predictions, selected reactors of each type were constructed. For the one-sided collector, f =16 mm and w =32 mm were set. For the two-sided collector, values of f =32 mm, d =32 mm, and w =38.4 mm were used. Finally, for the three-sided collector, a prototype with f =28.8 mm, d =32 mm, w =44.8 mm, and b =9.6 mm was built. The geometries of these three collectors were chosen within the ranges defined in Table 1 as simple, representative configurations of each family, with the sole purpose of experimentally validating the ray-tracing simulations rather than optimizing the f/D ratio or any other design parameter. For all of them, a collector length of 380 mm and an internal tube diameter of 13 mm (0.2 L of illuminated volume) were used. For the construction, reactors were modeled threedimensionally using CAD software. Subsequently, the collector support structure was printed in Nylon 12 using a selective laser sintering (SLS) 3D printer (Formlabs Fuse1) and coated with MIRO-SUN® material from ALANOD, with a reflectivity in the UV range of 90 %. The design was implemented to tilt the collection area 40º relative to the horizontal (corresponding to the local latitude of the Universidad Rey Juan Carlos facilities in M´ ostoles, Spain - 40.33◦N, 3.86◦W), as established in the simulations to achieve the highest solar concentration. On the other hand, for comparative purposes, a standard CPC Fig. 1. Schematic representation of one-, two-, and three-sided flat collectors and the parameters that define their geometries where: b is the bottom side of the 3sided collector (m), d the collector depth (m), D the diameter of the reactor tube (m), f the distance from the centre of the receiver tube to the collector (m), w the collector width (m), and β the angle between collector sides (º). Table 1 Configurations of the proposed collectors and simulations carried out. One-sided collector Two-sided collector Three-sided collector Minimum Maximum Interval Minimum Maximum Step Minimum Maximum Step f/D 1/2 3/2 1/10 7/10 3/2 1/5 3/10 3/2 3/10 w/D 1 3 1/5 1 3 1/5 1/2 3 1/2 d/D - - - 1/2 5/2 1/5 1/2 3/2 1/5 b/D - - - - - - 1 3 2/5 Nº reactors 121 605 1080 M. Martín-S´ omer et al. Journal of Environmental Chemical Engineering 14 (2026) 120718 3 collector made of high-quality anodized aluminum was used, ensuring 90 % direct light reflection and designed for the aforementioned tube dimensions. In Fig. 2, a photograph of the collectors studied experimentally can be observed. 2.4. Experimental test To validate the calculations derived from ray tracing simulations, potassium ferrioxalate actinometry experiments were conducted as described elsewhere [24] (2.47⋅mM initial concentration of Fe 3+ ). Additionally, bacterial inactivation assays were also performed using E. coli K12 bacteria (CECT 4624, corresponding to ATCC 23631, where CECT stands for ’Colecci´ on Espa˜ nola de Cultivos Tipo’) as the model microorganism. Fresh liquid cultures were prepared by inoculating into Luria-Bertani (LB) nutrient medium (Miller’s LB Broth, Scharlab) and incubating for 24 h at 37◦C with constant stirring on a rotary shaker. An initial bacterial concentration of 10 6 CFU mL −1 (0.9 % NaCl) was used in all experiments. Sample analysis was conducted throughout the reactions following a standard serial dilution procedure. Each decimal dilution was plated onto LB nutrient agar plates and then incubated at 37◦C for 24 h before counting. For all experimental reactions, the photoreactors were operated with a total volume of 2 L in recirculation mode with a reservoir tank. The flow rate was 12.3 L min −1 , powered by a centrifugal pump. Experiments under natural sunlight were conducted with the collectors facing south to maximize radiation collection. Incident radiation on the collector surfaces was measured using a PCE-UV34 radiometer. 3. Results 3.1. Experimental validation 3.1.1. Actinometrical tests The experimental validation was carried out by constructing one reactor of each type as described in Section 2.3. The solar concentration capabilities of one-, two-, and three-sided collectors were compared with that of a standard CPC collector with a 90º acceptance angle, as well as with the use of the reactor tube without any solar concentration system. The dynamic concentration factors for each of the chosen reactors, as well as the amount of solar radiation for each beam angle over a year at the facilities of Rey Juan Carlos University (considering an inclination of the reactors equal to the latitude and facing south) are shown in Fig. 3. Radiation with beam angles greater than 50 ◦is negligible, with most of it being below 25 ◦. Regarding the DCF values, the simple collectors reach values of around 70 % of those of standard CPC, with a notable similarity between the different reactors proposed. Furthermore, the one-sided collector, shows higher values for beam angles starting at 30 ◦ compared with the two-sided and three-sided collectors. This is relevant for locations where solar positions throughout the year often fall within this range of beam angles, as later discussed in relation to Fig. 7. Actinometric tests were conducted in each reactor to calculate the amount of concentrated radiation. Since the calculation of photons measured by actinometry depends on the quantum efficiency of the reaction, which varies with the wavelength of the received radiation, it is necessary to apply a quantum efficiency value not for a specific wavelength but for the entire range of wavelengths in the solar spectrum. This value was not found in the literature, so a methodology for its calculation was established, as shown in Eq. 1. Φsolar =∑Φλ⋅GHIλ⋅Φλ⋅κλ⋅Tλ⋅rλ ∑(GHIλ⋅Φλ⋅κλ⋅rλ)(1) Where GHI λ represents the spectral radiation energy reaching the Earth surface, measured as the vertical component of both direct and diffuse contributions [25], Φ λ is the quantum efficiency for each wavelength Fig. 2. Schematic representation and picture of the solar collector system. Fig. 3. DCF values for the different proposed reactors and solar dose over a year in Rey Juan Carlos University for each beam angle for actinometric reactions. M. Martín-S´ omer et al. Journal of Environmental Chemical Engineering 14 (2026) 120718 4 [26], κ λ is the calculated radiation absorption coefficient for each wavelength [27], T λ is the transmittance of the reactor tube for each wavelength, and r λ is the reflectivity of the collector material for each wavelength. T λ and r λ values were provided by the manufacturer. The final obtained value for the solar quantum efficiency in the reactor used was 1.092 E⋅mol⁻¹ . Reported values for the ferrioxalate actinometer include 1.21–1.26 at 365 nm and about 1.38 at 253.7 nm [28,29], showing that higher efficiencies are associated with shorter UV wavelengths [30]. Since the solar spectrum also contains a large fraction of longer wavelengths with lower quantum efficiencies, the obtained value is as a solar-weighted average that reasonably falls below those typically measured under monochromatic UV irradiation. Additionally, a similar approach was applied to calculate a solarweighted average wavelength (λsolar, nm) (Eq. 2) and a solar-specific absorption coefficient (κsolar∗, m 2 ⋅mol −1 ) (Eq. 3) which will be used in subsequent calculations. λsolar =∑λ⋅GHIλ⋅Φλ⋅κλ⋅Tλ⋅rλ ∑(GHIλ⋅Φλ⋅κλ⋅rλ)(2) The value obtained for λ solar was 399 nm. κsolar∗=∑κλ∗⋅GHIλ⋅Φλ⋅κλ⋅Tλ⋅rλ ∑(GHIλ⋅Φλ⋅κλ⋅rλ)(3) The value obtained for κsolar∗was 271 m²⋅mol⁻¹ . The κ λ * values were taken from [27]. In turn, the solar absorption coefficient (κ solar , m −1 ) was calculated by multiplying κ solar * by the initial 2.46 mol⋅m −3 Fe³ ⁺ concentration (C Fe3+ , mol⋅m −3 ). The resulting value of κsolar was 666.66 m −1 . Besides, a correction factor (Z) between the solar spectrum and the measurement range of the UV radiometer used (290–390 nm) was calculated according to Eq. 4, with an obtained value of 2.47. Z=∑(GHIλ⋅Φλ⋅κλ⋅rλ) ∑ 390 290 (GHIλ⋅Φλ⋅κλ⋅rλ) (4) Once these values were calculated, a geometric concentration ratio (GR, dimensionless) was determined for each reactor to compare their concentration capability as a function of geometry. Since the solar position varied during the actinometric experiments, GR was calculated as the slope of the linear representation between the radiative dose absorbed by the reactor tube determined by actinometry (D act , J⋅m −2 ) (Eq. 5) and the radiative dose incident in the normal direction to the collector surface (D ᵣₐ d , J⋅m⁻²) obtained from radiometer measurements (Eq. 6). Dact(t) = CFe2+(t)⋅Vsol Φsolar⋅Stube NAv⋅h⋅c λsolar (5) Where V sol is the reaction volume, C Fe 2+ is the concentration of Fe 2+ produced in the actinometric reaction, S tube is the external surface of the reactor tube, N Av is Avogadro’s number, h is Planck’s constant, c is the speed of light, λ solar is the representative solar wavelength calculated using Eq. 2, and t is the reaction time. Drad(t) = ∫t 0 qrad(t)⋅Z cos α (t)⋅sin(φ(t) − 90)⋅dt (6) Where q rad is the flux measured by radiometry on the opening surface of the collector (W⋅m⁻²), t is the reaction time (s), α is the angle between the direction of the incident radiation and the normal vector to the collector opening surface (rad), φ is the topocentric azimuth angle (eastward from north) (rad) and Z is the correction factor calculated according to Eq. 4. In Fig. 4a, it can be observed that GR increases from N =0 to CPC, reflecting the greater ability of more complex reflectors to redirect the incoming normal dose at the collector surface (D rad ) into the dose delivered to the tube (D ₐ ct ). Thus, higher GR denotes stronger geometric/optical concentration. Note that GR is a purely optical–geometric indicator and does not, by itself, account for practical trade-offs such as cost, projected area, or manufacturability. To validate the methodology used in the simulations, a validation factor (VF, dimensionless) was defined as the slope of the linear representation of D act versus the dose absorbed by the reactor tube derived from radiometer readings (D tube , J⋅m −2 ). D tube is computed introducing DCF in D rad calculation using Eq. 7. DCF term accounts for the ray path inside the collector and the reflective losses along it for each moment throughout the experiment’s progression, considering the solar position [23], while the second term converts the incident dose into in-tube dose. By comparing the two independent in-tube doses, D act and D tube , VF values close to 1 indicate a good fit, confirming the correctness of the conversion and validating the calculation method for subsequent use. Dtube(t) = ∫t 0 qrad(t)⋅Z⋅DCF(t) cos α (t)⋅sin(φ(t) − 90)⋅dt⋅Vtube⋅κsolar Stube (7) Where V tube is the tube volume, S tube is the external surface of the reactor tube, κ solar is the solar absorption coefficient, and the remaining magnitudes are explained in Eq. 6. The obtained values for VF are depicted in Fig. 4b. As previously discussed, a value of 1 confirms the validity of the procedure employed. It is evident from the figure that all calculated constants closely approximate 1, affirming the correctness of the methodology utilized. 3.1.2. Bacterial inactivation experiments To corroborate the validity of the predictions made for SODIS (optical density near 0), bacterial inactivation tests were carried out in each of the reactors. The UV dose received in each reactor was calculated Fig. 4. (a) Geometrical ratio and (b) validation factor for different collectors. M. Martín-S´ omer et al. Journal of Environmental Chemical Engineering 14 (2026) 120718 5 based on the measurements from the UV radiometer and the DCF of each reactor throughout the experiments, depending on the solar beam angle as shown in Eq. 8. Fig. 5 shows the DCF for proposed reactors for SODIS and solar radiation distribution over a year in Rey Juan Carlos University. DUV(t) = ∫t 0 qrad(t)⋅DCF(t) cos α (t)⋅sin(φ(t) − 90)⋅dt (8) It can be seen that no radiation with beam angles greater than 50 ◦is expected, with most of it being below 25 ◦. For beam angles below 25 ◦, the values obtained for the simple reactors are slightly lower than those for the standard CPC collector, but they are of the same order of magnitude and can therefore be considered comparable. If the DCF values are compared with those shown in Fig. 3, much higher values can be observed due to the lower absorption of radiation and therefore greater available incident radiation. On the other hand, Fig. 6 shows the inactivation results obtained for two experiments carried out in each of the reactors. It can be seen that, regardless of the reactor used, there is an excellent agreement between the inactivation achieved and the calculated dose. Therefore, these results validate the procedure followed also in the case of a process such as SODIS, allowing calculations and simulations to be carried out with complete guarantee. 3.2. Worldwide solar radiation The analysis of solar dose as a function of beam angle is essential for optimizing the design of solar collectors. However, evaluating each collector at every moment throughout the year would entail an extremely high computational cost. To address this limitation, a previously developed solar calculator [23]was used, allowing the determination of the yearly solar dose for each beam angle in a collector inclined at an angle relative to the horizontal equal to the latitude of the location. The results obtained, presented in Fig. 7 for a clear-sky scenario (0 % cloud cover), provide insight into how the angular distribution of dose varies with latitude. First, it is observed that locations closer to the equator exhibit the highest values of global dose, which is expected due to the greater availability of sunlight throughout the year. However, these regions also experience a greater variability in beam angles, meaning that the radiation is not concentrated in a narrow range but rather distributed over a broader spectrum of angles. This angular dispersion forces the use of more complex collectors that can capture radiation over a wider range of beam angles. Conversely, as latitude increases, the range of angles with significant radiation values decreases, suggesting that in regions farther from the equator, collectors can be optimized to capture radiation within a more defined and restricted angular interval. A notable feature of Fig. 7 is the presence of a common peak in dose around a beam angle of 25◦, observed across all latitudes. This occurs because the collector is inclined at an angle equal to the latitude of its location, which means that the sun’s rays strike the collector most efficiently at this particular beam angle. Since the inclination is chosen to optimize yearly solar capture, the system naturally aligns with the dominant solar position for that location, leading to a peak in dose at this angle of incidence. However, in intertropical regions (between 23.5◦N and 23.5◦S), an additional phenomenon is observed: there is a significant increase in solar dose for beam angles greater than 60◦. This is due to the way the solar position shifts throughout the year in these regions. Near the equator, the sun is not always in the southern part of the sky (as it is in higher latitudes); instead, it moves northward and southward over the course of the year. This means that at certain times, the solar noon position is nearly overhead or even behind the collector, which results in more radiation reaching the collector at very high beam angles. In contrast, in higher latitudes, the sun remains predominantly in one hemisphere, and such high beam angles are less frequent. On the other hand, Fig. 8 illustrates the impact of yearly cloud cover percentage on total solar dose, using the equator as an example. A Fig. 5. DCF values the different proposed reactors and solar dose over a year in Rey Juan Carlos University for each beam angle for SODIS experiments. Fig. 6. Bacterial inactivation in the different reactors based on the UV dose received in the reactor tube. Fig. 7. Yearly solar dose as a function of beam angle for different latitudes under clear-sky conditions (0 % cloud cover). M. Martín-S´ omer et al. Journal of Environmental Chemical Engineering 14 (2026) 120718 6 progressive reduction in available radiation is observed as cloud cover increases, indicating an inversely proportional relationship between the two factors. This behavior is attributed to the scattering and absorption of solar radiation by atmospheric particles, reducing the amount of direct radiation reaching the collector’s surface. This effect is particularly relevant in regions with high cloud cover, where diffuse radiation can represent a significant fraction of the available energy. In terms of design, these observations have direct implications for optimizing solar collectors. In equatorial regions, where radiation is abundant, but its angular distribution is broad, collectors must be designed to capture radiation over a wide range of beam angles. In contrast, at higher latitudes, where radiation is concentrated within a narrower interval, simpler collector geometries can be used, optimized to capture radiation at the predominant beam angle. Additionally, in areas with high cloud cover, collectors should prioritize capturing diffuse radiation, while in regions with predominantly clear skies, designs can focus on maximizing the collection of direct radiation. 3.3. One-sided collector DCF values for the studied flat-sided collectors were calculated using the ray tracing tool explained in Section 2.2 and combined with the results of the yearly solar dose shown in Figs. 7 and 8 using Eq. 9. The resulting annual energy was then divided by the tube cross-sectional area perpendicular to the beam to normalize the results to reactors of any size. Fig. 9 and figures S1-S3 show the yearly concentrated solar dose in each of the analyzed reactors, considering different latitudes and cloudiness levels. Only the configurations with the shortest focal distance have been represented, as these were the most efficient in all cases. DUV(t) = ∫t 0 TIR(t)⋅DCF(t)⋅dt (9) Where TIR, is the total incident radiation (TIR, W⋅m −2 ) computed using the solar calculator [23]. It is observed that, regardless of latitude, collectors with a higher w/ D ratio capture more radiation. This is because a greater collector width allows for the interception of a larger amount of incident radiation, thereby increasing the concentrated energy. However, the percentage increase in captured radiation decreases as the collector’s capture area increases, suggesting that there is a point of diminishing returns beyond which increasing the collector width provides only a marginal benefit. This aspect is fundamental in terms of cost-benefit analysis, as larger collectors may capture more energy but also require more material and space, which may increase manufacturing and installation costs. Comparing the cases w/D =3 and w/D =1, a ratio of 1.2 is obtained for latitude 0◦and 1.07 for latitude 50◦, regardless of cloudiness. This indicates that the effect of increased concentration with collector width is lower as the latitude increases. This is due to the fact that at higher latitudes, the range of incidence angles with high radiation is narrower, limiting the benefit of greater capture. On the other hand, the analysis of cloudiness effects shows an inversely proportional decrease in total concentrated radiation. As the percentage of cloudiness increases, the amount of captured radiation systematically decreases due to the scattering and absorption of solar radiation in the atmosphere. However, this impact is similar for all evaluated collectors, suggesting that differences in collector design do not significantly compensate for radiation loss due to cloudiness. Finally, it was observed that the standard CPC collector concentrates approximately twice as much radiation as the simple collectors. However, its greater complexity and manufacturing cost limit its feasibility in many applications. The simple collectors, although with lower concentration, offer an efficient and accessible alternative, with a design that is easier to manufacture and install. The difference in concentration compared to the standard CPC is not so significant, allowing these collectors to still capture a substantial amount of radiation. This makes them an attractive option for reducing costs and facilitating their implementation in a greater number of environments without significantly compromising radiation capture. Fig. 10 and figures S4-S6 present the results obtained when considering the projected area of the selected collector. For this purpose, the data obtained from Eq. 9 are used again; however, in this case the result is divided by the collector’s projected area on the horizontal plane (a value that varies with the collector design and the latitude of installation). Since the projected area is proportional to the tube dimensions, this quantity is likewise normalized and valid at any scale. Unlike the previous case, this case shows that the most efficient configurations are those with smaller w/D ratios. This occurs because, although increasing the collector width allows for greater radiation capture, the increase in projected area does not sufficiently compensate for the gain in concentrated energy when normalized per unit surface. One of the most relevant aspects is the variation in the trend of dose with latitude. In the previous analysis considering total radiation, increasing latitude led to a decrease in captured radiation due to a wider range of incidence angles. However, the opposite occurs in this case. When accounting for projected area, it becomes evident that collectors receive more radiation per unit of projected area at higher latitudes. This behavior is explained by the fact that as latitude increases, collectors become more inclined with respect to incoming radiation, reducing their projected area and thereby increasing the effective concentration of Fig. 8. Effect of cloud cover on yearly solar dose at the equator (0◦latitude). Fig. 9. Yearly concentrated solar dose for simple collectors with different w/D ratios and for a CPC with a 90◦acceptance angle, considering cloudiness percentage of 75 %. M. Martín-S´ omer et al. Journal of Environmental Chemical Engineering 14 (2026) 120718 7 radiation. A significant result is that collectors with w/D ratios up to 1.6 achieved higher doses per unit of projected area than the standard CPC collector. This indicates that, in terms of occupied surface, using simple collectors with moderate w/D ratios can be more efficient than employing a standard CPC, which, although more advanced in terms of concentration, requires a larger surface area. The choice of the most suitable collector ultimately depends on the design priorities. If the goal is to maximize total radiation capture, opting for larger w/D ratios is the best approach, as they allow for greater interception of incident radiation. On the other hand, if space efficiency is a key factor, collectors with lower w/D ratios offer a better solution, since their reduced projected area enables a more effective use of available space while still maintaining a substantial level of radiation capture. Another important criterion to consider is manufacturing costs, although they are difficult to quantify and were not included in this analysis. However, in any case, simple collectors, particularly those with lower w/D ratios, would always be more cost-effective than the standard CPC, making them a more accessible and practical choice in many applications. 3.4. Twoand three-sided collectors The analysis of the two and three-sided collectors was carried out in a similar manner to that previously discussed for the one-sided collector. Due to the large number of cases studied and the complexity of their analysis, it was decided to include only the optimal collectors in each case in the representations. Fig. 11 and figures S7-S9 show the yearly solar dose with the optimal collectors of one, two, and three flat sides, as well as the standard CPC and the bare tube. It can be observed that for latitudes located between the tropics, the two-sided and three-sided collectors barely improve the results obtained with the one-sided collector and are far from the efficiency obtained with the standard CPC. This is because, at these latitudes, the variation of the beam angle is very high (see Fig. 7), and the solar zenith varies between the south and the north, making it necessary to use collectors capable of concentrating radiation from any solar angle, such as the standard CPC or the onesided collector. At higher latitudes, a considerable increase in the radiation concentrated by the two-sided and three-sided collectors can be observed. For low cloudiness, the three-sided collector improves the concentration efficiency obtained with the standard CPC from latitudes of 35º, while the two-sided collector is capable of practically matching the standard CPC at a latitude of 50º. On the other hand, under high cloud cover conditions, the three-sided collector can still improve the efficiency of the CPC at high latitudes; however, its relative efficiency decreases due to the standard CPC’s great capacity to capture diffuse radiation (directly related to the degree of cloudiness). On the other hand, in Fig. 12 and figures S10-S12 it can be observed that when considering the projected area for all types of simple reactors studied, the efficiency obtained with the standard CPC is improved for any level of cloudiness considered (although the improvement decreases with the increase in cloudiness due to greater diffuse radiation). Additionally, for all reactors, according to this criterion an increase in efficiency was observed with the increase in latitude, which was attributed to the smaller projected area of the collectors when inclined to the latitude. In the case of using the bare tube, it was observed that for low latitudes, the efficiency was similar to that obtained by the standard CPC, although this similarity diminished for higher latitudes. 3.5. Overall performance assessment and comparison with previous studies The results obtained are particularly relevant, as they demonstrate that, for certain latitudes, a well-designed simple reactor can Fig. 10. Yearly concentrated dose/projected area for simple collectors with different w/D ratios and for a CPC with a 90◦acceptance angle, considering cloudiness percentage of 0 %. Fig. 11. Yearly concentrated solar dose for one-, twoand three-sided flat collectors and for a CPC with a 90◦acceptance angle, considering cloudiness percentages of 0 %. Fig. 12. Yearly concentrated solar dose/projected area for one-, twoand three-sided flat collectors and for a CPC with a 90◦acceptance angle, considering cloudiness percentage of 0 %. M. Martín-S´ omer et al. Journal of Environmental Chemical Engineering 14 (2026) 120718 8 concentrate solar radiation more effectively than a conventional CPC. These configurations not only offer a significantly lower cost due to their ease of construction but also show potential to match or even exceed the performance of standard CPC reactors, provided that the design is carefully optimized. The parameters defining the optimal configurations for each latitude and type of reactor, according to the two efficiency criteria analyzed, are detailed in the supplementary material (Tables S1 to S18). As can be observed in Fig. 11, N =1 collectors remain around 55 % of relative performance when compared to the CPC, while N =2 varies between 64 % and 98 %, and N =3 between 66 % and 113 %. When comparing these results with previous studies on simplified solar collectors, the relative optical performance aligns well with reported trends. Flat plate collectors typically achieve 55–65 % of the CPC 90◦ performance, which is consistent with the N =1 range observed here [31]. Linear Fresnel systems, considered among the most efficient simplified designs, can approach 85–95 %, overlapping with the upper range of N =2 and N =3 configurations [32]. Advanced CPC-based or hybrid geometries occasionally exceed the baseline CPC performance, as seen in the highest N =3 cases [33]. These are the reported values; however, in the cited studies, neither the variation with geographical location nor the projected area of the collectors is taken into account. As can be seen in Fig. 12, the designs optimized to maximize the dose-to-projected-area ratio substantially outperform the CPC. Similar trade-offs have been discussed in solar photo-Fenton research. Gomes et al. compared CPC with alternative geometries such as flat mirrors and double parabola reflectors and found that these simpler designs achieved comparable optical performance while considerably lowering fabrication costs and improving mechanical durability, especially when substituting anodized aluminum with more accessible materials like polished stainless steel [34]. Their findings highlight the importance of balancing optical optimization with implementation feasibility in low-resource settings. In parallel, Martínez-García et al. tested a pilot-scale solar disinfection system based on a V-trough reflector and reported that, despite receiving lower actinometric radiation than a CPC, it delivered a higher treatment volume per unit area [35]. This was attributed to the geometry’s favorable distribution of light and improved exposure of the reactor volume. This principle also explains the superior area efficiency observed in several of the simple geometries evaluated in the present work. Altogether, these results reinforce the idea that well-engineered simple designs can offer both functional and logistical advantages over CPC, especially in contexts where scalability, cost, and ease of deployment are critical. The results of this study contribute to ongoing research efforts aiming at optimizing solar reactor designs by considering not only their theoretical optical efficiency, but also their practical performance under real operating conditions. Although a full, site-specific cost analysis is beyond the scope of this study, we calculated indicative cost ranges for the collectors themselves using unit prices from a previous work for anodized aluminum reflective sheet and borosilicate tubing [20], machining and specialized fabrication costs have not been included because they are complex to estimate and vary widely across regions and time. For practical orientation, we report material-only cost ranges that cover the span from the cheapest to the most expensive among the optimal designs computed across the different latitudes considered: for the bare tube (N =0) approximately 10.55 € , for the one-sided flat collector (N =1) approximately 10.77 € -12.21 € , for the two-sided flat collector (N =2) approximately 11.07 € -14.25 € , and for the three-sided flat collector (N =3) approximately 10.77 € -13.43 € . These ranges are provided to give readers an order-of-magnitude idea of material costs across the set of optimal solutions; project-specific budgets should be obtained by combining the geometric dimensions given in the manuscript with current local prices and fabrication conditions. Building on the material-cost estimates described above, a preliminary evaluation of the treatment cost per cubic meter of water was carried out. This calculation was based on a target solar irradiance requirement of 40 W⋅m⁻², the concentration capacities obtained for each collector configuration and a ten-year amortization period for the material cost of each reactor. Using the yearly solar dose delivered by each optimal design, the corresponding annual treated volume per collector was estimated, and the amortized material cost was divided by this value to obtain an indicative treatment cost. Following this procedure, the resulting cost ranges extend from the cheapest to the most expensive among the optimal configurations identified across all latitudes. The estimated treatment costs fall within approximately 0.4031–0.6731 € ⋅m - ³ for the N =0 (bare tube) configuration, 0.2765–0.4234 € ⋅m - ³ for N=1, 0.2154–0.3533 € ⋅m - ³ for N =2, and 0.1721–0.3249 € ⋅m - ³ for N=3. If the treatment costs obtained here are compared with those of other low-cost water-treatment options reported in the literature, the advantages of the simplified flat-reflector collectors become apparent. PETbottle SODIS is effectively free in terms of energy input [36] (costs are limited to the occasional replacement of bottles), but it is constrained by low treatment capacity per unit, long exposure times under suboptimal weather, sensitivity to turbidity, and handling requirements, which limit its suitability for larger-volume or community-scale applications. Low-cost ceramic filters typically show very low per-liter costs (commonly <1 $⋅m⁻ 3 ) [37], and biosand filtration yields per-liter costs comparable to some concentrator systems; however, both technologies can present limitations in removing smaller pathogens such as certain viruses and may therefore require complementary treatments. Coagulant/chlorine sachet solutions generally incur higher per-liter expenses (>10 $⋅m⁻ 3 ). Reported treatment costs for CPC-based household reactors are on the order of 2 $⋅⋅m⁻³[38]. On the other hand, previous work [19] demonstrated that 3D printing can significantly reduce the manufacturing cost of solar collectors relative to commercial CPC units. However, when the treatment cost of the prototype developed in that study is calculated using the same methodology applied here, the resulting value is approximately 0.3772–0.6057 € ⋅m - ³ , which remains above the indicative treatment-cost obtained for the simplified flat-reflector collectors analysed in the present work. Consequently, while PET-bottle SODIS remains the lowest-cost option per liter, the simplified flat-reflector designs presented here offer substantially higher per-unit treatment capacity, improved scalability, and markedly lower material and construction complexity than many concentrator solutions, reinforcing their suitability for low-resource contexts where both affordability and treatment throughput are required. Beyond cost and optical performance, it is also relevant to place the operating conditions achievable with the proposed reactors in the context of multi-pathogen solar disinfection studies. In the present work, the results are expressed as yearly accumulated solar doses, but the concentration factors obtained for the simplified collectors are similar to those reported for CPCand V-trough-based solar reactors that have been successfully tested against protozoa and viral surrogates [39]. For example, static CPC systems and V-trough reactors operating under typical SODIS conditions have demonstrated effective inactivation of Cryptosporidium parvum and other protozoa, as well as several enteric viruses and bacteriophages, when exposed to sunlight over time scales of a few hours to one day [40]. The fact that the simplified geometries analysed here can deliver comparable solar doses over a year indicates that, under appropriate operating conditions, they could in principle be used within similar irradiance and temperature regimes to target a broader range of pathogens than E. coli alone. 4. Conclusions This study demonstrates that simple flat solar collectors can offer a low-cost and efficient alternative to conventional CPC collectors for M. Martín-S´ omer et al. Journal of Environmental Chemical Engineering 14 (2026) 120718 9