Solar water disinfection in large-volume containers for low-income countries

Currently, more than 800 million people lack access to safe drinking water. Among them, most live in low-income countries with limited financial and technological resources. In this context, the lack of centralised supply of drinking water produced in treatment plants is covered by processes at the household level. These treatments, usually referred to as Household Water Treatments (HWT), are characterised by being low-cost, user-friendly, and sustainable. Examples are chlorination, flocculation-coagulation, boiling, and solar water disinfection. The latter, also called SODIS process, has proved to be one of the most appropriate treatments for producing safe drinking water at the household level because it is inexpensive, not dependent on consumables, and has been widely demonstrated to be effective in removing pathogens (viruses, bacteria, and protozoa) from water. This process is based on the germicidal effect of UV light and its synergistic effect with rising water temperature. The procedure is very simple: just fill a transparent container with water and place it in sunlight for several hours. The most widely accepted procedure states that 2 L polyethylene terephthalate (PET) bottles should be exposed for 6 h on sunny days or 48 h on cloudy days. However, this standard method presents limitations that could be overcome by changing some features of the procedure: increasing the volume of the container would decrease the recontamination risk caused by handling several 2 L bottles and also maximise the production of safe drinking water; the use of container materials other than PET would significantly increase the efficacy of viruses and protozoa inactivation; and a detailed study of the factors influencing the process would reduce the overestimation of the solar exposure time.

Therefore, the development of accurate kinetic models is crucial to guarantee, and even maximise, the production of safe drinking water. An ideal kinetic model should consider all critical factors affecting the efficacy of the process, such as the intensity and spectral distribution of the solar radiation, the transmission spectrum of container walls —and the ageing of the SODIS container material— either made of PET or alternative plastic materials, the chemical composition of the water —as substances in the water that can play a critical role as radiation attenuators and/or sensitisers triggering inactivation processes— and the type and concentration of pathogens in the untreated water (viruses, protozoa, or bacteria).

In this context, the main objective of this PhD Thesis is to develop a methodology for the comprehensive kinetic modelling of the process that allows the accurate estimation of the solar exposure time required in large-volume SODIS containers. For this purpose, the entire model was divided into several parts, focused on the different factors discussed above:

▸development of an algorithm to easily obtain the actual solar daily dose anywhere in the world (solar radiation),

▸development of a calculation tool to estimate the spectral irradiance inside SODIS containers for the most suitable materials (container),

▸development of a procedure to calculate the effective incident radiation in large-volume SODIS containers as a function of the concentrations of naturally occurring substances found in natural waters (water composition), and

▸development of a set of mechanistic kinetic models to define the inactivation reactions of the different types of microbial targets during the SODIS process (kinetic modelling to inactivate viruses, protozoa, and bacteria).

The algorithm for calculating the daily dose depending on the day of the year and the latitude was developed taking into account: i) the day length and the irradiance of sunlight at solar noon, which only depend on the latitude and the day of the year, ii) a first correction factor (F₁) that relates the cumulated incident radiation to the daily dose if the solar irradiance corresponds to that at solar noon and is constant during the day length, and iii) a second correction factor (F₂) that relates the real and the maximum cumulated incident radiation. For F₂= 1 (maximum theoretical dose), F₁ was estimated as 0.43 at 307.5 nm and 45°N of latitude. The calculation was carried out by comparing predictions with the sum of theoretical hourly cumulated radiation values (obtained from the National Renewable Energy Laboratory) over a whole day. This value was validated with the cumulated radiation data for latitudes from 60°S to 60°N, with a step of 5°. As the value of F₂ depends on the weather conditions, it is strongly recommended to measure the actual irradiance in the field to determine F₂. However, if actual measurements are not available, a fair approximation of this value as a function of the latitude and longitude and based on historical data is presented graphically in this work. For example, the estimated F₂ = 0.58 for 52°N, 5°E, reduced the error between the predictions and actual measurements from 121% to 25.5%. The algorithm proved to be simple, accurate, and able to respond to different solar spectra. However, the simplicity of the procedure also involves a few limitations, such as the loss of accuracy for treatments shorter than half a day and for locations with widely varying weather, and the restricted use for latitudes outside the 60°S and 60°N gap.

The tool for calculating the radiation spectrum inside the SODIS containers was called Solar UV Calculator and was developed following the Lambert-Beer Law and measuring the absorption spectra of previously selected plastic materials. The first selection was made based on literature data of production costs, mechanical properties, and photostability, and experimental measurements of optical properties. Initially, polystyrene (PS), polyvinyl chloride (PVC), and polyethylene (PE) were discarded as suitable materials for the manufacture of SODIS containers because of their poor photostability. Although polypropylene (PP) also presents poor photostability, it was chosen due to its very low cost, its transmission of radiation in the UVB range, and the possibility of adding a UV-stabiliser. Polycarbonate (PC) and polymethylmethacrylate (PMMA) were also selected due to their great photostability and good optical properties. Finally, despite its opacity in the UVB range, PET was also chosen to compare the standard plastic used in SODIS containers with the above-mentioned alternative plastic materials. Therefore, the Solar UV Calculator tool offers default radiation spectra for PMMA, PET, PC, and PP. Nevertheless, this tool is freely available to any potential user interested in the evaluation of SODIS containers design parameters (thickness and other types of materials if the absorption spectrum is known), or even in the evaluation of other solar processes subjected to a strong spectral dependence on the transmission of the materials. In addition, an exhaustive study of the effect of weathering on the properties of the plastic used in SODIS processes currently employed in the field (PET, PMMA, PP, and PP with 1% by weight of UV-stabiliser) was also carried out. For this purpose, mechanical properties (flexural and tensile tests), physical and chemical properties (infrared spectroscopy, differential scanning calorimetry and high temperature gel permeation chromatography analyses), optical properties (UV-Vis spectrophotometry) and bacterial disinfection rates (experiments under controlled illumination conditions) of accelerated ageing-plastic samples were analysed. PET and PP showed the lowest disinfection rates due to the fact that PET does not transmit UVB radiation, and PP suffered significant ageing and, consequently, its transmittance dropped. The lifetime of PET was estimated, at least, as 1 year of solar exposure, while that of the PP without additives was only 2 months. PP with 1% UV-stabiliser and PMMA showed the best optical properties and disinfection rates despite ageing, with no signs of significant degradation after 9 months for PP+1% and, at least, 1 year for PMMA. PP+1% presents a good balance between elasticity and impact resistance and, therefore, it is recommended for its use in portable SODIS devices, while PMMA is recommended for static SODIS devices due to its high stiffness, flexural strength, and its ease of scratching. Thus, PMMA and PP with a 1% UV-stabiliser were identified as optimal materials for manufacturing SODIS devices, what confirms that ideal mechanical properties are not crucial for selecting suitable materials and the need to study weathering SODIS containers.

The effective available radiation inside large-volume SODIS containers was calculated considering the concentration of naturally occurring substances in water. Disinfection experiments with increasing concentrations of the species in water were carried out. The substances studied were (bi)carbonates, soluble carbohydrates measured as dissolved organic carbon (DOC), iron, humic acids, and solids measured as turbidity. UV-transparent substances such as bicarbonates and soluble carbohydrates had no impact on the disinfection rate, while the presence of optically active substances such as iron, humic acids, or solids acted as radiation attenuators. The optical properties of the water (absorption and scattering coefficients and phase function of the scattering) with the optically active substances were calculated by spectrophotometry. The distribution of radiation within the large-volume containers was estimated by numerical simulation. The absorbing substances (iron and humic acids) led to a progressive decrease of incident radiation along the container as the radiation pathway length increased. In contrast, the scattering particles (turbidity) led to significantly more pronounced profiles with much higher values of incident radiation close to the front side and a uniformity index significantly lower. Further disinfection experiments with optically active substances were performed to prove that the kinetics of the process in large-volume containers is not only affected by the average value of the incident radiation but also by the homogeneity in the radiation distribution. This hypothesis was successfully validated for all substances except iron, which also plays a role as an enhancer of bacterial damage, probably because of its possible permeability within the cell and its contribution to the intracellular Fenton process.

For mechanistic kinetic modelling, experimental disinfection data were obtained under controlled light and water temperature conditions for viruses, protozoa, and bacteria. These data helped to study each of the mechanisms (thermal, photonic, and synergistic UV-T synergistic), propose appropriate kinetic schemes for each microbe inactivation, and estimate the corresponding kinetic parameters of the model. Those parameters that were not available in the literature or could not be estimated independently were calculated using regression models.

For the modelling of virus inactivation by the SODIS process, MS2 viruses showed no thermal effect under dark conditions in the range of water temperatures usually reached during the SODIS process (20-50°C). Experimental data of virus inactivation under illumination conditions at low temperatures showed a linear dependence that was successfully modelled with a first-order kinetic model. In contrast, viruses did show sensitivity to UV-T synergistic effect (3-log reduction achieved in 60-90 min), which was modelled by applying a modified Arrhenius equation to the kinetic constant of the photonic reaction. In addition, the spectral dependence was included in the photonic reaction by adding the biological weighting function of RNA to its kinetic constant. Further experiments —the radiation emission spectrum was modified by placing PET, PP, and PMMA plastics between the water and the solar source— were performed and successfully responded to the RNA spectral action. These results confirmed the negligible inactivation observed experimentally for PET, the most commonly used material for SODIS processes, and also the great disinfection rates for PMMA and PP.

For the modelling of protozoa inactivation by the SODIS process, the system was also first analysed under dark conditions. Under these conditions, Cryptosporidium parvum protozoa showed thermal inactivation above 30°C (modelled with the Arrhenius equation), being especially significant above 40°C (3-log reduction achieved in 17 h at 44°C and in approximately 1 h at 50°C). Experimental disinfection curves under illumination conditions at low temperatures showed an initial shoulder that was successfully modelled with a series-event model. In addition, the protozoa also showed sensitivity to the UV-T synergistic effect (3-log reduction achieved in 2-3 h), which was modelled by applying an Arrhenius equation to the kinetic constant of the photonic reaction. Spectral dependence was also included in the photonic reaction by adding the DNA biological weighting function (calculated previously) to its kinetic constant. Further experiments —the radiation emission spectrum was modified by placing PET, PP, and PMMA plastics between the water and the solar source— were performed and successfully responded to the DNA spectral action. These results confirmed the negligible inactivation observed experimentally for PET at temperatures below 40°C. Consequently, the null inactivation of viruses and the exclusively thermal inactivation of protozoa in PET containers are evidence of the need to look for new alternative materials, such as PMMA or PP with UV-stabiliser.

For the modelling of the inactivation of bacteria by the SODIS process, experiments with H₂O₂ added to the water matrix were also performed to elucidate the internal cellular mechanisms and their kinetic parameters. First, the cellular respiration pathways and the effect of radicals’ damage on the inactivation of Escherichia coli bacteria was modelled (dark conditions). For the latter, a series-event model with a recovery constant was used to accurately reproduce this system. The behaviour of bacteria when they are exposed to H₂O₂ and its permeation into the cell altering its equilibrium was also considered. In addition, the H₂O₂ sinks —thermal decomposition, permeation into the cell, interaction with the cell membrane and with cell debris— were also defined kinetically. Bacteria also showed thermal inactivation (under dark conditions) above 30°C, which was modelled with the Arrhenius equation. Finally, photonic effects were included in the mechanisms (deactivation/activation of enzymes/coenzymes, intracellular photo-Fenton process, and direct damage). Direct damage was modelled with another series-event model with recovery that was coupled with that of radicals’ damage using a multiple hit-multiple target model. In addition, bacteria showed a soft sensitivity to UV-T synergy and was modelled by adding the Arrhenius equation to the kinetic constant of direct photonic damage.

In conclusion, this PhD Thesis presents the development of a comprehensive kinetic model that estimates the required solar exposure time in large-volume containers subjected to the SODIS process. The complexity of the microbes forces the adoption of assumptions to get plausible mechanistic kinetic models. However, despite their limitations, these models offer an unparalleled framework for future developments and improvements, and new reactions, processes, and microbial targets can be envisaged.