Capstone Project Sample: Designing a Low-Cost Household Water-Filtration Unit

Model Answer

Abstract

Access to safe drinking water remains a critical global health challenge, particularly in low-resource and disaster-affected settings where centralised treatment infrastructure is unavailable or has failed. This capstone project reports the design, construction and laboratory evaluation of a low-cost household water-filtration unit built from locally obtainable materials, intended to reduce turbidity and microbial contamination in untreated surface water towards World Health Organization guideline ranges. Three prototype iterations were developed using a layered filtration approach combining coarse gravel, river sand, granular activated carbon and a ceramic candle filter element, with each iteration tested against turbidity, indicator-organism reduction and flow-rate criteria. The final design achieved a turbidity reduction of 96%, an Escherichia coli log reduction of 3.1, and a sustained flow rate of 2.4 litres per hour, at an estimated unit material cost of £11.40. The project demonstrates that a household-scale unit combining low-cost, locally available materials can achieve meaningful water-quality improvement, while highlighting practical limitations around filter-media replacement, viral contaminants, and the need for field validation beyond laboratory conditions.

Introduction and Problem Statement

An estimated two billion people worldwide use a drinking-water source contaminated with faecal matter, and unsafe water remains a leading contributor to diarrhoeal disease burden, particularly among children under five in low- and middle-income countries (World Health Organization, 2022; Bain et al., 2014). Where centralised treatment and distribution infrastructure is absent, damaged, or interrupted, for example following flooding or in informal settlements, household-scale point-of-use treatment can provide an interim or supplementary barrier against waterborne pathogens and reduce diarrhoeal disease incidence (Sobsey et al., 2008; Fewtrell and Colford, 2004). Existing commercial point-of-use filters are often effective but too costly for the populations most affected, creating a case for low-cost, locally buildable alternatives.

This capstone project addresses the problem: can a household water-filtration unit, built primarily from low-cost and locally obtainable materials, achieve a meaningful reduction in turbidity and microbial contamination that approaches World Health Organization drinking-water guideline values? The project’s aim was to design, build and laboratory-test such a unit, with the following objectives: (1) to achieve a turbidity reduction of at least 90% from a simulated untreated surface-water source; (2) to achieve at least a 3-log reduction in an indicator organism; (3) to sustain a flow rate adequate for a household of four to five people (approximately 20 litres per day); and (4) to keep total unit material cost below £15.

These objectives were chosen to reflect the practical constraints most often reported as barriers to household filter adoption: performance sufficient to make a genuine difference to health outcomes, a flow rate fast enough that a family will actually use the unit daily rather than reverting to an untreated source out of impatience, and a cost low enough to be replicable by a non-governmental organisation distributing units at scale or by a household purchasing its own unit outright. The design brief also specified, informally, that the unit should require no electricity and no imported proprietary cartridges, so that it could be assembled, and where possible repaired, using materials available through general hardware or aquarium-supply retailers in most urban and peri-urban settings.

The scope of the project was bounded to bench-scale prototyping and laboratory testing using simulated turbid water rather than field deployment, and to microbial indicator testing using a non-pathogenic surrogate organism rather than pathogenic strains, in line with the laboratory’s safety protocols. Chemical contaminants, such as heavy metals and agricultural run-off residues, were outside the project’s scope, as was any formal cost-effectiveness or willingness-to-pay analysis, which would require field data on target-population income and existing water-treatment behaviour beyond what a laboratory-based capstone could reasonably collect.

Background and Literature

Household water treatment technologies are typically evaluated against the World Health Organization’s performance criteria for household water treatment options, which classify technologies as achieving ‘highly protective’, ‘protective’ or ‘interim’ status based on log-reduction values for bacteria, viruses and protozoa (World Health Organization, 2021; Hunter, 2009). Ceramic filtration, in particular, has an established evidence base: ceramic pot and candle filters, often silver-impregnated to inhibit biofilm growth, have been shown in multiple field and laboratory studies to achieve bacterial log reductions of 2 to 4, though virus removal is typically weaker due to pore sizes larger than viral particles (Brown and Sobsey, 2010; Lantagne, 2001).

Biosand and slow sand filtration, meanwhile, relies on the formation of a biological layer (the ‘schmutzdecke’) at the sand surface, which can achieve significant pathogen reduction once matured over several weeks, but performance is variable during the initial ripening period and sensitive to flow rate and maintenance (CAWST, 2012; Mellor et al., 2013). This ripening dependency was a factor in the decision not to rely on biosand filtration alone for this project: a maturation period of several weeks would have been difficult to accommodate within the capstone’s testing timeline, and immature biosand layers can, in the interim, offer little better than mechanical straining, which risked producing misleadingly poor early-stage results unrelated to the design’s eventual steady-state performance. Granular activated carbon is widely used to address taste, odour and some organic contaminants but is not, on its own, an effective microbial barrier, meaning it is typically deployed as one stage within a multi-barrier design rather than as a standalone treatment (Sobsey et al., 2008).

The multi-barrier principle, combining several treatment mechanisms so that the overall system’s protection does not depend on any single stage, is well established in both engineering and public-health water-treatment literature and directly informed this project’s layered design, which combines physical straining (gravel and sand), adsorption (activated carbon) and a ceramic microfiltration barrier (candle element) in series (Davis, 2010). Unlike higher-cost membrane-based decentralised systems (Peter-Varbanets et al., 2009), cost has consistently been identified as the principal barrier to household filter adoption in low-income settings, reinforcing the rationale for prioritising low-cost, locally sourceable materials over higher-performing but more expensive commercial cartridge systems (Clasen, 2009).

A further recurring theme in the literature is the gap between laboratory and field performance: several evaluations of household filters that performed strongly on the bench have shown reduced effectiveness once deployed in real households, attributed variously to inconsistent maintenance, incorrect assembly, and source water more variable than typical laboratory test suspensions (Hunter, 2009). This gap directly shaped the framing of this project’s own limitations, discussed later, and reinforced the decision to treat the laboratory results reported here as a necessary first stage of validation rather than as evidence of field-ready performance.

Approach and Methodology

The project followed an iterative engineering design process: define requirements, design a candidate solution, build a prototype, test against defined criteria, and refine before the next iteration. Design requirements were derived directly from the project objectives (turbidity reduction, microbial log reduction, flow rate, cost) and supplemented by a practical constraint that all components should be sourceable from general hardware or aquarium-supply retailers, to reflect the project’s low-resource design brief.

The unit’s layered design comprised, from top to bottom within a 20-litre food-grade plastic vessel: a coarse gravel pre-filter layer (approximately 5 cm) to remove large sediment and protect lower layers; a river sand layer (approximately 15 cm) for further particulate removal; a granular activated carbon layer (approximately 8 cm) for adsorption of dissolved organics and residual taste and odour compounds; and a silver-impregnated ceramic candle filter element as the final microbial barrier, feeding into a lower collection vessel by gravity flow. All media were sourced from general aquarium-supply and hardware retailers and rinsed thoroughly with clean water before assembly to remove fine dust that could otherwise cause premature clouding of the filtrate during the first draw-through, a step identified during initial pilot testing as necessary after an early trial run produced visibly turbid output on first use despite the media itself performing well on subsequent draws.

Three prototype iterations were built and tested. Prototype 1 used the layered media without the ceramic candle element, to establish a baseline for the physical/adsorptive stages alone. Prototype 2 added the ceramic candle element. Prototype 3 refined the gravel and sand layer proportions following Prototype 2’s results, and added a simple flow-restrictor collar to slow flow through the ceramic element, improving contact time. A final design consolidated the changes from Prototype 3 with a minor adjustment to carbon layer depth. Each prototype was housed in an identical outer vessel so that differences in measured performance could be attributed to the internal media configuration rather than to variation in vessel geometry, and each was allowed a short settling period after assembly before testing began, to allow any residual fine media dust to clear from the system.

Testing used a simulated turbid water source prepared in the laboratory by adding a standard kaolin clay suspension and a non-pathogenic Escherichia coli surrogate strain to tap water, following a protocol adapted from published household water-treatment testing methods (Sobsey et al., 2008). Turbidity was measured before and after filtration using a bench nephelometer and reported in Nephelometric Turbidity Units; microbial reduction was measured by membrane filtration and colony counting following standard enumeration methods (British Standards Institution, 2014), expressed as a log10 reduction value; flow rate was measured as the time taken to collect one litre of filtrate under a standard head height. Each prototype was tested in triplicate and mean values reported. Unit material cost was calculated from itemised component prices at the time of build. Ethical approval was not required as the project involved no human participants and used a non-pathogenic laboratory strain under standard laboratory biosafety protocols, confirmed with the department’s laboratory safety officer prior to testing.

The influent (untreated) test water was standardised at a target turbidity of approximately 45 to 55 NTU for every test run, chosen to represent a moderately turbid surface-water source such as a slow-flowing river after rainfall, rather than an extreme worst-case suspension, so that results would be broadly representative of typical field conditions rather than an artificially easy or artificially difficult test. The E. coli surrogate strain was seeded at a standardised concentration confirmed by plate count immediately before each test run, and control samples of the influent water were tested alongside every prototype run to confirm that the seeded concentration and turbidity were consistent across the testing programme, allowing performance differences between prototypes to be attributed to the design changes rather than to variation in the challenge water itself.

Implementation / Findings

Prototype 1, without the ceramic candle element, achieved a turbidity reduction of 55% and negligible microbial log reduction (0.4), confirming that the gravel, sand and carbon layers alone provided meaningful particulate removal but were insufficient as a microbial barrier. Prototype 2, with the ceramic candle element added, achieved a substantially improved turbidity reduction of 72% and a microbial log reduction of 2.1, though flow rate fell to 1.1 litres per hour, below the project’s practical usability threshold.

Prototype 3, with revised media proportions and the flow-restrictor collar to increase ceramic-element contact time, achieved a turbidity reduction of 89% and a microbial log reduction of 2.8, with flow rate recovering to 2.0 litres per hour. The final design, incorporating a small increase in carbon layer depth, achieved the project’s best results: a turbidity reduction of 96%, a microbial log reduction of 3.1, and a sustained flow rate of 2.4 litres per hour, meeting all four of the project’s original design objectives, including the cost target of an estimated £11.40 per unit. Table 1 summarises performance across all four iterations, and Figure 1 presents the turbidity-reduction trend.

The clearest single design lesson from the testing programme was that the flow-restrictor collar introduced in Prototype 3 was disproportionately influential: it recovered most of the flow rate lost in Prototype 2 while also increasing removal performance, because slower flow through the ceramic candle increased contact time without any change in media. This finding was not anticipated from the design’s initial engineering calculations, which had focused on media depth rather than flow control, and illustrates the practical value of iterative empirical testing over simulation alone.

Evaluation of Outcomes

Measured against the four original design objectives, the final prototype met or exceeded every target: turbidity reduction of 96% against a 90% target, a microbial log reduction of 3.1 against a 3-log target, a flow rate of 2.4 litres per hour comfortably supporting the 20-litres-per-day household target, and a unit material cost of £11.40 against a £15 ceiling. Classified against the World Health Organization’s household water-treatment performance criteria, the final design’s microbial log reduction places it within the ‘protective’ category for bacteria, though it does not achieve the higher log-reduction values required for a ‘highly protective’ rating, and virus removal was not directly tested, a known limitation of ceramic-based filtration noted in the literature review (Brown and Sobsey, 2010).

The project’s principal engineering success was identifying that flow control, rather than media selection alone, was the critical variable separating Prototype 2 from Prototype 3’s performance, a finding that would not have emerged from a purely calculation-based design process and that has direct implications for future iterations. The principal limitation is that testing was conducted entirely under controlled laboratory conditions using a simulated water source and a non-pathogenic surrogate organism; real-world source water is more variable in turbidity, organic content and pathogen load, and field performance cannot be assumed to match laboratory results without further validation (CAWST, 2012).

A further limitation concerns filter-media longevity: the testing programme evaluated performance only for a single filtration cycle per prototype and did not assess how turbidity reduction or microbial log reduction change as the ceramic element becomes fouled over repeated use, which is a critical determinant of a household filter’s real-world protective value.

It is also worth noting that the four design iterations were not tested against identical, independently randomised challenge water on every occasion; while control samples confirmed broadly consistent influent turbidity and seeded organism concentration across runs, minor batch-to-batch variation in the kaolin suspension cannot be entirely excluded as a contributing factor to the measured improvements between prototypes, alongside the intended design changes. A larger number of replicate runs per prototype, beyond the triplicate testing used here, would strengthen confidence that the observed step-changes in performance reflect the design modifications themselves rather than test-water variability.

Recommendations and Reflection

Three recommendations follow from the project. First, before any field deployment, the final design should undergo an extended-use trial, cleaning and replacing the ceramic candle element on a defined schedule, to establish a realistic media-replacement interval and to confirm that performance does not degrade unacceptably between cleanings. Second, field testing using an authentic local source water, rather than a laboratory-simulated suspension, should be conducted to validate whether laboratory turbidity and log-reduction figures hold under real-world conditions, in line with the limitations identified above. Third, given that ceramic filtration alone does not reliably address viral contamination, a low-cost supplementary disinfection step, such as a solar disinfection stage or chlorine dosing, should be evaluated as an addition to the design for contexts where viral risk is a significant concern, in line with established humanitarian water-supply guidance (Sphere Association, 2018; Reed, 2011) and low-cost technical guidance for household filtration (Practical Action, 2016).

As a piece of applied engineering work, the project’s principal strength was its iterative, test-driven design process, which allowed an unanticipated but highly consequential design variable, flow control, to be identified empirically rather than assumed from first-principles calculation alone. This reinforced, in practice, the value the literature places on iterative prototyping over single-pass design for appropriate-technology water treatment (Davis, 2010).

The clearest limitation, on reflection, was scope: the decision to test in laboratory rather than field conditions, made for practical reasons of time and safety approval, means the project can report promising bench-scale performance but cannot yet claim real-world effectiveness. A future iteration of this capstone would benefit from partnering with a public-health or humanitarian-engineering organisation to conduct field trials, and from extending the testing protocol to include repeated-use and viral-surrogate testing, both of which were beyond the resource and time constraints of the present project.

Finally, the project underscored a broader lesson about appropriate-technology engineering: the most valuable design change identified here, the flow-restrictor collar, cost pence to add and required no specialist manufacturing capability, yet delivered a disproportionate improvement in both flow rate and removal performance once it was tested rather than merely calculated. This suggests that future household-filtration design work in resource-constrained settings should place at least as much weight on cheap, testable, incremental modifications as on higher-cost material substitutions, since the former are both more affordable to iterate on and, on this evidence, potentially just as consequential to overall performance.

References

Bain, R., Cronk, R., Wright, J., Yang, H., Slaymaker, T. and Bartram, J. (2014) ‘Fecal contamination of drinking-water in low- and middle-income countries: a systematic review and meta-analysis’, PLOS Medicine, 11(5), e1001644.

British Standards Institution (2014) BS EN ISO 9308-1: Water Quality — Enumeration of Escherichia coli and Coliform Bacteria. London: BSI.

Brown, J. and Sobsey, M.D. (2010) ‘Microbiological effectiveness of locally produced ceramic filters for drinking water treatment in Cambodia’, Journal of Water and Health, 8(1), pp. 1–10.

CAWST (Centre for Affordable Water and Sanitation Technology) (2012) Biosand Filter Manual: Design, Construction, Installation, Operation and Maintenance. Calgary: CAWST.

Clasen, T. (2009) ‘Scaling up household water treatment: looking back, seeing forward’, Journal of Water and Health, 7(4), pp. 1–8.

Davis, M.L. (2010) Water and Wastewater Engineering: Design Principles and Practice. New York: McGraw-Hill.

Fewtrell, L. and Colford, J.M. (2004) Water, Sanitation and Hygiene: Interventions and Diarrhoea. Washington, DC: World Bank.

Hunter, P.R. (2009) ‘Household water treatment in developing countries: comparing different intervention types using meta-regression’, Environmental Science & Technology, 43(23), pp. 8991–8997.

Lantagne, D.S. (2001) Investigation of the Potters for Peace Colloidal Silver Impregnated Ceramic Filter. Boston: Alethia Environmental.

Mellor, J.E., Smith, J.A., Learmonth, G.P. and Netshiongolwe, N. (2013) ‘Modelling the effectiveness of household water treatment’, Water Research, 47(2), pp. 519–528.

Peter-Varbanets, M., Zurbrügg, C., Swartz, C. and Pronk, W. (2009) ‘Decentralized systems for potable water and the potential of membrane technology’, Water Research, 43(2), pp. 245–265.

Practical Action (2016) Household Water Treatment: Technical Brief. Rugby: Practical Action.

Reed, B. (2011) Sustainable Water and Sanitation for Emergencies and Disasters. Loughborough: WEDC.

Sobsey, M.D., Stauber, C.E., Casanova, L.M., Brown, J.M. and Elliott, M.A. (2008) ‘Point of use household drinking water filtration: a practical, effective solution for providing sustained access to safe drinking water in the developing world’, Environmental Science & Technology, 42(12), pp. 4261–4267.

Sphere Association (2018) The Sphere Handbook: Humanitarian Charter and Minimum Standards in Humanitarian Response. Geneva: Sphere Association.

World Health Organization (2021) Results of Round II of the WHO International Scheme to Evaluate Household Water Treatment Technologies. Geneva: WHO.

World Health Organization (2022) Guidelines for Drinking-Water Quality. 4th edn. Geneva: WHO.