Design and evaluation of zero-energy UVC-LED reactor fitted with hand pump system for disinfection

India has experienced an enormous growth in population and mainly relies on the use of rainwater harvesting for drinking purposes. India, being a monsoon country, receives rainfall for only three to four months in a year, but it is not evenly distributed in all places. As rainwater harvesting is deeply rooted in South Indian history, so-called ‘Ooranies’ or large water ponds are still likely to be found all over the rural areas. There may be the existence of pathogenic contamination which requires an advanced level of treatment for these water sources before consumption. Disinfection is one of the most effective methods to reduce the number of waterborne diseases caused by pathogenic organisms. Availability of clean and safe drinking water at the doorstep of the ordinary person is still a need of the hour.

‘Ooranies’/large water ponds are one of the main sources to cover the drinking water demand in villages in South India (IWWA 2020). They collect the rainfall runoff and preserve it as a drinking water source for the village throughout the year. Hence, it is an annual storage facility. In general, oorani water is found to have a high turbidity and bacterial count. The pre-treatment (sand filtration) applied removes the turbidity, but often an appropriate disinfection step is lacking in the drinking water system of South Indian villages. Hand pump is the major drinking water point-of-use for the communities in the villages. Out of the total hand pump users (996 million), 80% of them are from rural populations (UNICEF 2012).

Chorine-based disinfection is the common method for removal of pathogenic organisms in drinking water, usually the final treatment before consumption (Yoo 2010). Different disinfection methods known so far are: boiling, bio-sand filtration, chlorination, ozonation, UV and solar photolysis (Sobsey et al. 2008). Process efficiency of UV treatment is high, but the major drawbacks are the presence of mercury content, the cost of electrical energy and frequent replacement of the conventional UV mercury lamp which has a lower life period (Chatterley & Linden 2010). Many works have focused on the use of UV-LEDs for disinfection, which is energy efficient and free of toxic substances (Song et al. 2016; Li et al. 2019). One of the disadvantages of UV disinfection is the revival of microorganisms in darkness after treatment, whereas higher UV intensity and lower exposure time minimises the microorganisms revival in darkness (Chen et al. 2020). On the other hand, cost and energy efficiency are to be considered as well. Energy efficiency might be achieved by optimising the UV dose per log inactivation with varying wavelength. Wavelength of 260 nm could effectively absorb DNA molecules resulting in pyrimidine dimers, pyrimidine hydrates and crosslinks between proteins and DNA, but it was reported by Takahashi et al. (2020) that the microorganisms could dark repair their DNA. In contrast, it was reported that at 280 nm, protein absorption was at its peak which leads to disruption of enzymes that favour dark revival (Song et al. 2016). Also, an important challenge for the UV-LED disinfection technique is the cost factor, especially among low-income groups. Therefore, an economically feasible and self-sustaining disinfection system is required, which is the reason wavelength ∼280 nm was chosen for the present study.

In the present study, Pattikadu village in Kanchipuram district (Tamil Nadu, South India) was chosen as the study area. The Oorani water (raw water) was found to be highly turbid (∼50 NTU) and with the presence of Escherichia coli bacteria (∼2,400 CFU/mL). It was observed that after sand filtration, turbidity (∼2 NTU) significantly reduced, but the biological parameter (∼133 CFU/mL) did not comply with the drinking water quality standards (BIS 10500:2012). Therefore, disinfection prior to consumption of water was mandatory, for which, zero-energy technology for removal of pathogenic organisms was designed. UVC-LED disinfection was evaluated for the developed annular photo-reactor which was fitted with the hand pump of Pattikadu village (study area). The mechanical energy was conserved and stored as electrical energy to serve as the power supply for energising the UVC-LEDs. The developed model would be zero-energy technology for practical applicability in rural areas.

Water samples from the hand pump were taken from Pattikadu village (Figure 1, adopted from Google map images and Figure 2), located in Tirukalukundram Block of Kancheepuram district in Tamil Nadu, South India. The total geographical area is around 294 hectares with a total population of 1,050 people (Wikivillage 2020). The pre-treatment of Pattikadu oorani consists of a multi-barrier system (Figure 3, adopted from Government of Tamil Nadu Rural Development for Rejuvenation of Ooranies). The multi-barrier system contains a catchment area, supply channel, grit chamber, oorani storage, filter chamber, water supply pipe, fine sand filter and draw well. The grit chamber with a baffle wall traps the grit particles and prevents floating solids entering the catchment. Oorani serves as an aerobic pond where microbiological processes take place under aerobic conditions. In the horizontal roughing filter, larger organics and inorganics are removed. Further water is pumped to a fine-sand filter removing turbidity and coliforms and withdrawn from the hand pump for use. The raw water was collected from the water spread area (Oorani) (No. 4 in Figure 3). The fine sand-filtered water was collected from the draw well (No. 6 in Figure 3).

Each 10 L of raw and sand-filtered water samples before and after monsoon rains were collected and analysed within 48 hours of collection for accurate results. Table 1 reports the date and time of sample collection.

The UVC-LEDs with nominal peak emission at 275 ± 5 nm that were adopted by previous literature survey demonstration proved that the highest E. coli inactivation occurred at 275 nm rather than 255 or 310 nm (Aoyagi et al. 2011; Oguma et al. 2013). The emission spectrum was determined by fibre optic spectrometers (StellarNet EPP 2000) and by SpectraWiz. The values are derived from the source and plotted as a graph between intensity and wavelength. The maximum intensity count was observed at 275 nm from the plotted graph (Figure 4). Each UVC-LED was operated at 20 mA constant forward current, 5.5 V forward voltage, viewing angle 120° and radiant flux of 0.0937 mW for each LED, according to the manufacturers’ details.

An annular UVC-LED photo-reactor (capacity 1 litre) was designed for batch process. 24 UVC-LEDs were connected in parallel and placed inside a quartz tube of diameter 30 mm and length 300 mm. The quartz tube was fitted inside an enclosure made of stainless steel of diameter 87 mm and length 240 mm and thickness 2 mm. The distance between the light source and nearer/farther side of the water medium was 1.5 and 2.75 cm, respectively. A schematic diagram of the UVC-LED reactor and lamp image is shown in Figure 5.

The hand lever of hand pump Mark II was welded with the necessary support angles at appropriate places to achieve the full rotation of the circular plate, where the plate rotates with the mechanical up and down movement of the hand lever (Niroopa & Kanmani 2015). The circular plate was connected to the large gear wheel which was connected to the dynamo through a small gear wheel. One rotation of the larger gear wheel induces the smaller gear to rotate ten numbers of revolutions. On average, in 1 minute, 60 strokes (up and down movement of lever) can be generated by a person. The mechanical energy of the gear wheel was converted into electrical energy through a dynamo. The dynamo consists of stator coils and rotor magnet, generating AC current through Faraday's law of induction. The dynamo was connected to the 6 V DC battery through a bridge rectifier. The voltage generated by the hand pump lever produced 600 revolutions in the small gear wheel per minute, which was enough to charge a 6 V battery. The experimental setup of the hand pump is shown in Figures 6 and 7.

In general, an average person produces 60–80 strokes in a hand pump for a minute. Therefore, when the energy balance is computed, the force applied to the hand lever creates the rotational energy in the larger gear wheel. The ratio between the radius of the larger wheel and smaller wheel is 10. The number of revolutions increases to 600 in the smaller gear wheel. The rotations of rotor magnet in the dynamo cause a change in magnetic flux. The current is induced in the stator coil by Faraday's law of induction by change of magnetic flux. Electromotive force is energy generated per unit charge. A multi-meter recorded a minimum of 6 V and 20 mA when the hand pump was operated for each stroke. The temperature difference (ΔT) of water before and after disinfection was recorded to be 1.5 °C for 6 minutes of reactor operation.

pH was measured by pH meter according to IS 3025 (Part 11) 1983 (RA 2012). The instrument was standardised with two buffer solutions of pH near that of the sample and a different pH value. Temperature was measured and adjusted accordingly. The electrode was washed and immersed in the sample beaker to record the sample pH.

Water temperature was measured by digital thermometer according to IS 3025 (Part 9) 1984. This was done by dipping the thermometer into the sample and recording a stable reading.

Turbidity was measured by a turbidimeter according to IS 3025 (Part 10) 1984 (RA 2017). First, the instrument was checked for accuracy in calibration with standards. Then, the sample was shaken to achieve uniform dispersion and poured onto the turbidimeter tube to record the turbidity value.

Electrical conductivity was measured using a pre-calibrated digital conductivity meter. The probe was dipped into the water samples until a stable reading was obtained.

The pour plate method was followed according to IS 14648: 2011. In sterile Petri dishes, 1 mL of the sample was poured into 20 mL of nutrient agar medium. The suspension was mixed with medium by tilting the plates slowly. After solidification, inoculated plates were inverted and incubated at 30 °C for 3 days and colonies were counted. Serial dilution was done until 30–300 colonies were observed in a single plate for quantitative estimation and all experiments were performed six times independently including the duplicate plates.

Parameters were analysed for the collected raw water and sand-filtered water before and after monsoon. Table 2 shows the water quality parameters for raw and sand-filtered water before and after monsoon. It was observed that turbidity and microbial concentration is comparatively low before monsoon, but after rain, the turbidity, total suspended solids and microbial concentration values increased, whereas total dissolved solids and electrical conductivity showed a slight increase due to dissolution of inorganic ions because of rainwater surface run-off. All the parameters were analysed three times independently. The values reported are means with standard deviations of the values for the respective parameters. The sand-filtered water (after monsoon) was used for disinfection studies in the present work.

The disinfection efficiency is shown in Table 3 and Figure 8, where the initial colonies count was 133 ± 8 CFU/mL. With the exposure to UVC-LED irradiance, the colony counts reduced significantly with time resulting in <1 CFU/mL. The standard plate count values reported are means with standard deviations of the three repeatable values.

From Figure 9, the value of k (0.57 cm²/mJ) was obtained from the slope of linear regression models (R² = 0.97). The sensitivity of bacteria to UVC radiation was evaluated by UVC dose-based inactivation rate constant (slope of log inactivation and UVC dose). The obtained k value is comparable with past studies (Chatterley & Linden 2010), where the k for E. coli inactivation varied from 0.17 to 0.422 cm²/mJ.

The germicidal efficiency of UV disinfection is majorly influenced by the emission wavelength but not the spectral sensitivity of target organisms (Chen et al. 2009). Even though the peak absorption of E. coli is around 260–265 nm, 275 nm proved to be efficient in terms of UV dose per log inactivation (Bowker et al. 2011). So far, UV disinfection works have been carried out using different wavelengths (315–400 nm (UVA), 280–315 nm (UVB) and <280 nm (UVC)). In general, at higher wavelength (365 nm or 310 nm), the UV dose-response required was higher due to the higher values of output energy of UVA and UVB-LEDs. On the other hand, lower wavelength (255 nm or 275 nm) requires least UV dose-response due to lesser output energy of UVC-LEDs. Hamamoto et al. (2007) demonstrated a higher value of 55,263 mJ/cm² UV dose was required for per log inactivation, where the emission wavelength was 365 nm. Similar work was carried out by Mori et al. (2007), who applied 365 nm for E. coli inactivation which required 13,846 mJ/cm² as the UV dose per log inactivation (Song et al. 2016). Bowker et al. (2011) applied 255 and 275 nm radiation for 1 log E. coli inactivation which required 3.3 and 2.4 mJ/cm². In this work, 2.23 mJ/cm² of UVC dose was required per log inactivation which proves to be more efficient than other works have reported so far. Further 2 log inactivation was obtained at 6 minutes under UVC dosage of 4.68 mJ/cm², and similar results were found in the literature (Chang et al. 1985; Vilhunen et al. 2009). Shorter disinfection time and higher UV dose is desirable for E. coli inactivation due to photochemical and biological processes (Harm 1980).

In this present work, the UVC-LED (λemission 275 ± 5 nm) reactor for disinfection was designed for rural areas, where hand pump is the major point-of-use for the small community population. The electrical power required for the disinfection was taken from the mechanical energy from up and down movement of the hand pump lever. The 6 V rechargeable battery was charged by 1 minute operation of the hand pump (60–80 strokes). The UVC-LED batch annular reactor (capacity 1 L) fitted with the hand pump showed that in 6 minutes of disinfection time, 100% E. coli inactivation was observed. The UVC dose-based inactivation rate constant obtained was 0.57 cm²/mJ. The survival UVC dose graph shows that the two-stage decay model is the best model fit for the present work. The novelty of the study is the use of high-end water treatment technology for lower-income groups with zero-energy requirement. In future, the reactor may be evaluated for higher throughput with other UV disinfection methods, including UVA/UVC combined treatment, UVA/TiO₂ photo-catalytic disinfection or pulsed UV disinfection.

This work was financially supported by the Department of Drinking Water and Sanitation, Government of India and Centre for Technology Development and Transfer, Anna University, Chennai.

All relevant data are included in the paper or its Supplementary Information.