Project Amazi · 2026
Four years of research in Gakenke District, a first system installed in Minazi Sector, and a business case for Rwanda’s water sector. Engineered at Imperial. Built in Rwanda.
01 · The problem
That is what communities in Gakenke District told us, again and again, across four years of interviews and wide-scale surveys.
Daily water use · Minazi Sector
3 to 20litres per person per day
Short of basic access, and far below the 50 L that intermediate access, the level that supports good hygiene, calls for.
litres per person per day
How rural households use water
Most household water doesn’t need to be drinkable. Rain can supply it.
Landslides · Gakenke District
The human cost
The national cost
Sources: e.quinox surveys and interviews in Gakenke, 2022 to 2025; WHO (2020) Domestic water quantity, service level and health; Dusabeyezu & Nsanzumukiza (2024); e.quinox 2026 partnership brief.
02 · Why it isn’t solved
Rain is abundant and the benefits of harvesting it are well understood. Yet adoption is largely confined to public buildings.
Bottleneck 1
There are no subsidies, so the whole cost lands on the families who need water most.
Amazi is 3.2× cheaper to install than a pumped tap connection. See the cost
Bottleneck 2
Existing tanks are large and heavy. Villages away from the main roads are hard, or impossible, to reach.
Amazi breaks down into parts one person can carry, cutting delivery costs by 90%. See the system
03 · The system
Each 700 L system links several tanks in parallel, and families add tanks as their needs and means allow. Every part is chosen for clean water, security and easy maintenance.
Metal-sheet roofs feed a gutter and downpipe. Guttering sets the yield, so it is budgeted, not an afterthought.
Emptied after each dry period, so the dirtiest water goes to irrigation, not the tank.
Stainless-steel mosquito mesh: unreactive, durable, and simple to inspect, clean and replace.
Water enters gently, without stirring up sediment that has settled.
Joined by wire-reinforced, tamper-proof pipe into one theft-proof system that fills faster and overflows less.
Meshed air vent on every tank. Lid clasps take a padlock, and lids lift off for cleaning.
Fills a 20 L jerrycan directly.
Excess water drains into gravel, so there is no standing water.
Cement, metal or wood, to suit each site.
Built to reach every village
−90%
Delivery costs 10,000 RWF per household, against 100,000 RWF for existing 1,000 L tanks. Components travel by moto and public transport, and can move with a family if they are relocated.
Sourced in Rwanda
Tanks are bought in Kigali, from a Tanzanian manufacturer. Pipes and guttering are made in Rwanda or imported from neighbouring countries, and Rwandan manufacturers are exploring custom orders.
04 · The cost
A private Amazi supply for one household in Gakenke, compared with a communal pumped tap connection from WASAC.
3.2×
cheaper to install680,000 RWF against 2.2M RWF per household
3.8×
cheaper to runmonthly operating cost per household
Amazi costs the same as today’s harvesters, but spends it differently. Delivery falls from 100,000 to 10,000 RWF and foundations from 160,000 to 60,000, while more goes on the filters and fittings that keep water clean.
Breakdown (RWF)
| Component | Amazi 700 L | Existing 1,000 L |
|---|---|---|
| Tanks | 180,000 | 160,000 |
| Delivery | 10,000 | 100,000 |
| Foundation materials | 60,000 | 160,000 |
| Accessories | 300,000 | 130,000 |
| Guttering | 130,000 | 130,000 |
| Total | 680,000 | 680,000 |
Monthly operating cost (RWF)
No pumps, no electricity and no watchman. Running costs are mostly cleaning and occasional parts, which families can handle with minimal tools.
Not accounting for economies of scale, visitor pricing, or under-reporting of WASAC costs. The average Rwandan household is 4 people. Sources: e.quinox procurement data, 2024 and 2026; WASAC Gakenke, 2024.
05 · The yield
Modelled on 1991 to 2020 monthly rainfall for each province, at 50 L per person per day and a conservative 50% capture rate. Metal-sheet roofs typically capture 85 to 90%.
% of intermediate-access demand · 50% capture
Hover or tap the chart to read a month. Only June to August, the long dry season, fall below 40%.
Where it works
The Eastern Province gets the least rain, yet still averages 42% of needs met, and it is home to over a quarter of Rwanda’s people. Our first system is in Gakenke, in the Northern Province.
| Province | Needs met (year avg) | Share of population | Rain per month |
|---|---|---|---|
| Western | 60% | 21.9% | 117 mm |
| Southern | 54% | 22.7% | 105 mm |
| Northern | 51.5% | 15.4% | 100 mm |
| Kigali | 44.5% | 13.2% | 87 mm |
| Eastern | 42% | 26.9% | 82 mm |
Population: Fifth Rwanda Population and Housing Census, 2022. Rainfall: Climate Change Knowledge Portal, 1991 to 2021.
06 · Water quality
Samples were tested in the University of Rwanda’s chemistry labs and compared with Rwanda’s rural drinking-water quality framework.
| Parameter | Amazi rainwater after first flush |
Tank uncleaned for 7 months |
WASAC tap Kigali |
Standard treated water |
Standard natural water |
|---|---|---|---|---|---|
| TurbidityFNU | 1.81within | 2.89 | 0.46 | ≤ 5 | ≤ 25 |
| pH | 5.83low | 5.56 | 7.30 | 6.5 to 8.5 | 5.5 to 9.5 |
| Total dissolved solidsmg/L | 14.32within | 7.73 | 46.1 | ≤ 700 | ≤ 1,500 |
| ConductivityµS/cm | 30.7within | 16.47 | 94.7 | ≤ 1,500 | ≤ 2,500 |
Untreated rainwater is suitable for
Still being studied
Our samples were slightly acidic. Their pH sits below the treated-water range but inside the natural-water range, and we are still investigating what its low pH and conductivity mean in practice.
07 · Beyond water
Knock-on effects the business case identifies, still to be quantified.
For the network
Lower demand means lower pressure, fewer leaks, slower wear and less pumping energy. Stored water also cushions families when piped supply is intermittent.
For the climate
Rainfall is growing more intense and less predictable, raising landslide risk in the Northern and Western provinces. Household storage cuts peak runoff, and the system moves with families relocated from high-risk zones.
08 · The timeline
First visit to Gakenke. Communities named everyday water access as their biggest limit.
Wide-scale surveys in Minazi Sector. Over 100 families and leaders met across both visits.
The modular 700 L system was designed at Imperial and its Rwandan supply chain mapped.
Installed in Minazi to test deployability, supply chain and yield. The business case was presented in April.
A multi-household pilot, and a place for rainwater harvesting in Rwanda’s water masterplans.
09 · The partnership
Working with
Business case lead: Riccardo Casonato, Imperial College London.
Who needs to be at the table
Organisations the business case identifies for bringing rainwater harvesting into Rwanda’s water supply.
Get involved
The first system is in. The next step is a pilot across more households, and that takes partners.
Partners & funders
Sponsors, NGOs, research groups and funding bodies: help take Amazi from one system to many.
Get in touchSupport a household
680,000 RWF covers the tanks, guttering, fittings, foundation and delivery for one household’s 700 L system.
DonateImperial students
Engineers, researchers and organisers from any department. The committee recruits every year.
How we work