A Low-Cost Two-Stage UV + Activated-Charcoal Filter for E. coli Removal in Household Drinking Water: Design, Bench Evaluation, and 30-Day Trial

Abstract

We designed, built, and evaluated a two-stage household water filter that combines a coconut-shell activated-charcoal pre-filter with a 6 W low-pressure UV-C exposure chamber. The prototype was assembled from off-the-shelf plumbing components at a bill-of-materials cost of ₹1,140. Across 90 controlled trials at three turbidity levels (5, 20, 50 NTU) using E. coliK-12–spiked feed water at 10⁶ CFU/100 mL, the two-stage design achieved log₁₀ reduction values (LRV) of 5.1 ± 0.2, 4.6 ± 0.3, and 3.4 ± 0.4 respectively — outperforming a UV-only control by a mean of 1.4 log at 50 NTU (p< 0.001, two-way ANOVA). One unplanned failure on day 12 (lamp intensity dropped to 62% of nominal) revealed a real-world risk of silent under-treatment; we propose a low-cost UV-indicator card as a mitigation. The design meets the WHO “highly protective” threshold (≥4 LRV) up to 20 NTU.

1. Introduction

Access to microbiologically safe drinking water remains uneven in rural India. The UNICEF/WHO Joint Monitoring Programme (2023) estimates that ~200 million people in the country rely on sources that intermittently exceed the WHO limit for E. coli(0 CFU/100 mL). Point-of-use (POU) treatment shifts the responsibility for the final microbiological barrier to the household, which is often the most reliable place for it, given distribution-network contamination.

Commercial UV-C purifiers are effective but retail for ₹8,000–₹15,000, placing them out of reach for the households who need them most. This paper investigates whether a sub-₹1,200 two-stage design can close that gap without sacrificing microbiological performance under realistic turbidity.

2. Literature Review

UV-C at 254 nm inactivates E. coliat doses above ~30 mJ/cm² (Hijnen et al., 2006). Turbidity attenuates UV transmittance approximately according to Beer–Lambert; at turbidities above ~20 NTU, achievable LRV degrades sharply (LeChevallier & Au, 2004). Activated carbon adsorbs dissolved organics and reduces UV absorbance, but on its own achieves only marginal (~0.5) LRV against bacteria (Sobsey, 2002).

Several published designs have combined pre-filtration with UV, but almost all target laboratory or municipal-scale flow, and none we found meets all four of the constraints listed in Section 3.1. This paper contributes an empirical evaluation of a design that does.

3. Methodology

3.1 Design constraints

• Bill-of-materials cost < ₹1,200 at single-unit quantities.
• No moving parts (excluding a peristaltic pump used only for testing; see §3.3).
• Runs on 12 V (compatible with a 20 W solar panel + 7 Ah battery).
• Delivers 2 L/min throughput to match household draw.

3.2 Build

The pre-filter is a 50 mm PVC housing (150 mm long) packed with 200 g of coconut-shell activated carbon (12×40 mesh), retained between two 25 µm nylon disks. The UV chamber is a 300 mm stainless-steel tube with a 6 W Philips TUV UV-C lamp (λ = 254 nm) mounted axially in a quartz sleeve. Full BOM, drawings, and photos are in Appendix A.

3.3 Test protocol

Feed water was dechlorinated tap water spiked with E. coliK-12 to a target concentration of 10⁶ CFU/100 mL, verified by pre-treatment sampling. Turbidity was adjusted with sterile kaolin to 5, 20, or 50 NTU (measured on a Hach 2100Q). Three configurations were tested: (a) two-stage; (b) UV-only (charcoal cartridge removed); (c) charcoal-only (UV lamp powered off). Each configuration × turbidity combination was run for 30 consecutive days (one trial per day). A peristaltic pump held flow at 2.0 ± 0.05 L/min.

Samples of 100 mL were collected pre- and post-treatment, filtered through 0.45 µm membranes, and cultured on m-ColiBlue24 media at 35 °C for 24 h. Blue colonies were counted as E. coli(Method 10029, Hach).

3.4 Statistical analysis

Log₁₀ reduction values were computed per trial as log₁₀(C_in / C_out). Data were analysed by two-way ANOVA with configuration and turbidity as factors, and a Tukey HSD post-hoc comparison. All analyses were performed in R 4.4.

4. Results

Mean LRV (± 1 SD) across the 30-day trial for each condition are shown below.

The two-stage configuration outperformed UV-only by a mean of 0.4 log at 5 NTU, 1.5 log at 20 NTU, and 1.4 log at 50 NTU. The interaction term (configuration × turbidity) was significant (F(4, 261) = 38.2, p< 0.001), confirming that the charcoal stage’s benefit grows with turbidity.

4.1 The day-12 lamp incident

On day 12 of the trial, a routine intensity check with a Solarmeter Model 8.0 UV-C sensor showed the lamp emitting only 62% of nominal (measured at the sleeve surface). Post-treatment CFU counts on that day were correspondingly elevated (LRV dropped to 2.1 at 20 NTU vs. 4.6 mean). No visible change in the lamp or filter had prompted the check. Root cause: internal quartz sleeve fouling from precipitated calcium. This trial was excluded from the mean LRV analysis but is discussed in Section 5.

5. Discussion

The results support the central hypothesis: at moderate-to-high turbidity (≥20 NTU), the two-stage design substantially outperforms UV-only, and the interaction is statistically significant. At 50 NTU, however, even the two-stage design falls below the WHO “highly protective” threshold. For source waters routinely above 50 NTU, additional pre-treatment (sedimentation or a finer particulate filter) would be needed.

The day-12 lamp incident is arguably the most important finding of this paper for real-world deployment. A user with no way to check UV-C intensity would have continued drinking water at <99% inactivation for days. We propose a passive UV-C indicator card (a photo-reactive strip that darkens irreversibly on adequate exposure) mounted inside the viewing port; this addition adds ~₹40 to the BOM and gives the user a daily visual go/no-go signal.

6. Conclusion

A two-stage charcoal + UV-C filter can be built for ₹1,140 in parts, achieves ≥4 LRV against E. coli up to 20 NTU, and outperforms UV-only by ~1.5 log at higher turbidity. The primary real-world risk is silent lamp degradation, mitigated by a low-cost passive indicator. Future work should extend testing to environmental E. coli isolates, add a virus surrogate (MS2 bacteriophage), and replace the peristaltic pump with a gravity-fed head tank suitable for household use.

Acknowledgements

Thanks to the biology lab at my school for autoclave/incubator access; to Dr. R. Iyer at the municipal water testing lab for guidance on membrane-filter technique; and to my physics teacher for lending the Solarmeter Model 8.0 that caught the day-12 incident.

References

1. Hijnen, W. A. M., Beerendonk, E. F., & Medema, G. J. (2006). Inactivation credit of UV radiation for viruses, bacteria and protozoan (oo)cysts in water: A review. Water Research, 40(1), 3–22.
2. LeChevallier, M. W., & Au, K.-K. (2004). Water Treatment and Pathogen Control. IWA Publishing, London.
3. Sobsey, M. D. (2002). Managing Water in the Home: Accelerated Health Gains from Improved Water Supply. WHO.
4. WHO (2017). Guidelines for Drinking-water Quality, 4th ed.
5. UNICEF/WHO Joint Monitoring Programme (2023). Progress on household drinking water, sanitation and hygiene 2000–2022.

Appendix A — Bill of Materials