← Água CidadãLer em Português →
Published: 2026-07-13

The State of the Almada Water System

A look at what the public numbers actually say — and why the current crisis was entirely foreseeable.

In the summer of 2026, taps ran dry across Almada. Costa da Caparica, Sobreda, the Charneca, Laranjeiro, Feijó — households went hours without water, pressure collapsed at night, and cistern trucks appeared in the worst-hit streets. The official framing has been one of exception: a heatwave, a surge in seasonal population, a "pressure without precedent" on the network. All of that is true. None of it is the real story.

The real story is written in the utility's own published figures, and it has been legible for years. The heatwave didn't break the system. It removed the last bit of slack from a system that was already running on empty.

The system, in numbers

(The figures in this section are the utility's own published system data, dated February 2024.)

Almada's water comes almost entirely from underground. Thirty-two boreholes — 93% of them physically located in the neighbouring Seixal municipality — feed a network serving about 177,000 people. The infrastructure is substantial: roughly 797 km of distribution mains, another 79 km of trunk adductors, 9 pumping stations, and 25 reservoirs offering 42 storage cells. On an average day, around 46,000 m³ of water enters distribution. Maximum daily production tops out near 57,500 m³.

That last pair of numbers is where the trouble starts. Average demand already sits at 80% of the absolute production ceiling. There is very little headroom — and a large, permanent leak steadily eating into what headroom exists.

The leak

Divide daily input by population and you get about 260 litres per person per day entering the network. But a lot of that never reaches anyone. According to the regulator, ERSAR, losses run at roughly 33% of water produced; the municipality's own figure is about 35% of input going unbilled. Strip out the water that's legitimately used but not charged for, and you're left with real, physical losses — water leaking into the ground — of something on the order of 11,000 m³ every day.

To judge whether that's bad, you can't just look at a percentage. The fair measure the water industry uses is the Infrastructure Leakage Index (ILI): actual losses divided by the theoretical minimum losses that even a perfectly maintained network of that exact size and pressure would suffer. A well-run utility sits between 1 and 2. Private Portuguese operators average close to that. The Portuguese public-utility average is worse.

Run Almada's numbers through the standard formula and the network lands at an ILI of around 5 — losing roughly five times the unavoidable minimum, or about 250 litres per connection per day. ERSAR's own historical rating for this is blunt: insatisfatória. Unsatisfactory. And this is not a new verdict — the same "insatisfatória" label was attached to Almada's losses back in 2018, at almost exactly the same per-connection figure (259 litres/connection/day).

Why it isn't fixing itself

Leaks in an ageing network don't fade away; they compound. Old pipe leaks more, breaks more often, and each break spawns new leaks. The only sustainable cure is replacing pipe faster than it decays. Here is the number that explains everything:

Over the past decade, Almada renewed its mains at an average rate of 0.42% per year.

At that pace, the network replaces itself roughly once every 240 years. The engineering guidance for a network of this age is somewhere between 1.5% and 4% per year. Almada has been renewing its pipes at a small fraction of the rate needed simply to stop the situation getting worse — never mind improving it. The infrastructure has been ageing considerably faster than it has been repaired, every single year, for years.

So when the heatwave arrived and demand pushed up against that 57,500 m³ ceiling, there was no margin to absorb it. Around 24% of daily production was — and still is — leaking into the subsoil. The emergency measures tell the story precisely: reduce pressure across the whole council between midnight and 6 a.m. so the reservoirs can refill overnight. Cutting pressure genuinely does reduce leakage — leakage rises faster than pressure, so easing off buys a more-than-proportional saving — but as a permanent nightly ritual it is a symptom being managed, not a problem being solved.

What could actually change this

The tools are neither exotic nor expensive, and the utility already has the bones for them — that dense grid of 42 reservoir cells and 9 pumping stations is close to ideal for zoning the network into metered districts.

There is one honest catch. Meters in Almada are still read manually, once a quarter. That, more than any algorithm, is the real ceiling on how smart this system can be — you cannot balance the books in real time on quarterly data. The analytics are cheap and well understood. The sensing layer that would feed them is the actual gap, and the actual investment.

Getting the ILI from about 5 down toward 2 would recover something like 6,000–7,000 m³ a day — comparable to commissioning a whole new borehole field, except this water is already paid for, already pumped, and already treated. It is simply being lost.

The uncomfortable question

None of the above is a revelation. The losses were rated unsatisfactory in 2018. The renewal rate has been a fraction of what's needed for a decade. The regulator has flagged it, reports have documented it, and the physics has never been in dispute. The current crisis didn't introduce a new problem — it exposed an old one that the summer heat simply made impossible to ignore.

So here is the question worth sitting with, addressed to whoever is willing to answer it:

If this has been known, measured, and named for so many years — why is not enough happening?

Sources and references

System data (February 2024 baseline)

Losses and network condition

Methodological note: the Infrastructure Leakage Index (ILI ≈ 5), the ~11,000 m³/day real-loss estimate, the UARL floor, and the 6,000–7,000 m³/day recovery potential are the author's own derivations from the published figures above, using the standard IWA water-balance methodology. They are estimates, not audited figures, and depend on assumptions about connection count and average network pressure.