What a water purifier really costs to run

Of the thirteen water purifiers here, the six sold in India run from ₹2,449 to ₹26,900, median ₹15,999; the seven sold in China run from ¥1,499 to ¥4,599, median ¥2,099. The two sets are not conversions of one another, and no machine here is sold in both markets — read whichever column applies to you and ignore the other.

Whichever column that is, the price is the part of the cost you can see. Twelve of the thirteen rows are reverse osmosis, and an RO machine consumes things: pre-filters two or three times a year, a membrane every few years, a post-carbon cartridge, and — the one nobody puts on the price tag — several litres of mains water for every litre it hands you. On the model further down, three years of that comes to roughly a third of everything you will have spent.

Two kinds of number are mixed below. Catalogue figures — prices, the 1–3 upkeep level, the eco flag, flow and storage — are exact. Service prices, filter-change intervals and waste-water ratios are not in the catalogue at all, so those come as ranges and as a labelled model, because the honest answer depends on what your supply is doing to the machine.

An RO machine is a queue of stages, and the queue exists because each stage protects the next. The order is most of what you need to predict the bill. Do you need reverse osmosis? covers whether you want the queue at all.

1. 1

   Sediment pre-filter — the cheapest part, changed most often

   A spun or pleated polypropylene cartridge, usually rated around 5 microns, whose whole job is to stop silt, rust flakes and pipe scale reaching anything downstream. Manufacturers typically schedule it at six to twelve months, but the interval is set by your supply: on a borewell, or where the mains runs muddy after monsoon, it can load up in three.

2. 2

   Carbon pre-filter — the membrane’s bodyguard

   Activated carbon, granular or as a block, adsorbing chlorine, chloramine and the organics that make water taste of the pipe. Removing chlorine is a requirement, not a refinement — see the next section. Same six-to-twelve-month range, but note that carbon is genuinely consumed rather than merely clogged: its adsorption capacity is finite, and when it runs out it stops working without looking any different.

3. 3

   The RO membrane — the expensive one

   A spiral-wound thin-film composite element, rated in gallons per day, and by a wide margin the costliest consumable. Two to three years is the usual expectation on Indian municipal water — less on high-TDS groundwater, more on a well-protected machine fed soft city water. Some Chinese machines quote longer; Midea’s Baize 800G advertises five years. Treat any such figure as a claim made under favourable conditions.

4. 4

   Post-carbon and mineraliser — taste, and sometimes minerals

   RO permeate is stripped of dissolved solids and tastes flat, so most machines polish it afterwards. Six of the thirteen rows carry a mineraliser, including the Kent Grand Plus, Pureit Copper+ and Aquaguard Aura copper-dosing designs sold in India. Consumables too, typically annual, and the stage people forget exists until the water starts tasting of nothing.

5. 5

   The UV lamp, where there is one

   Four of the six Indian rows name UV in the model itself — Kent Grand Plus RO+UV+UF, Pureit Copper+ Eco RO+UV+MF, Aquaguard Aura RO+UV+UF and Livpure Glo RO+UV. A lamp does not clog; it dims. Output falls with hours run, so lamps go on a calendar, commonly annual, even though the machine gives no sign anything changed. A lamp left in for four years is a decorative blue light.

Nothing in that list is calendar-driven by nature. Cartridges are consumed by litres and by load, so a household of two on clean municipal supply and one of six on borewell water can own the same machine on entirely different schedules — and the second will find out the hard way if it follows the manual.

Here is the single most expensive error available in this category, and it is committed by people trying to save money. A thin-film composite RO membrane is a polyamide layer a fraction of a micron thick, and polyamide is chemically intolerant of free chlorine: oxidation attacks the amide linkages, the rejection layer loses its selectivity, and the damage is cumulative and permanent. It cannot be cleaned out. The membrane keeps passing water — often more than before, which is why nobody notices — while letting more and more dissolved solids through with it.

The only thing standing between chlorinated municipal water and that membrane is the carbon pre-filter, at a small fraction of the price. Once the carbon bed is exhausted it stops adsorbing, chlorine passes straight through, and the clock starts on an element several times its cost. An exhausted carbon cartridge looks exactly like a fresh one.

The sediment stage fails differently and just as expensively. As it clogs it strangles flow into the pump, and a starved pump makes less permeate for the same feed — recovery falls, and the reject stream gets more concentrated than the design allows. Push a hard supply far enough and calcium carbonate and sulphate precipitate onto the membrane as scale. Scaling is at least partly reversible with cleaning; chlorine damage is not.

Twelve of the thirteen rows are rated to handle high-TDS supply; the exception is the gravity filter, which does not touch dissolved solids at all. But "handles high TDS" means the machine is built for that duty. It does not mean the membrane is indifferent to it.

Every water row carries a `maintenanceCostLevel` from 1 to 3. It is a band rather than a currency figure, and it is deliberately relative: a 1 in the Chinese list and a 1 in the Indian list both mean "cheap for their market". Across the thirteen rows the distribution is five at level 1, six at level 2 and two at level 3, median 2.

Upkeep level tracks complexity far more closely than it tracks price. The cheapest machine in the catalogue and two of the most sophisticated Chinese tankless units share level 1, because a short filter queue is cheap to refill whether it cost ₹2,449 or ¥2,299. The two level-3 machines are the most expensive rows in their respective markets, for the same reason twice: more stages, more specialised parts.

If running cost is your priority, the water finder acts on it directly. Choose "cheapest to maintain" at the wishes question and a level-1 machine gains twelve points while a level-3 machine loses eight — a twenty-point swing in a hundred-point score, usually enough to reorder the shortlist. It is a thumb on the scale rather than a filter, so a level-2 machine that suits you in every other respect can still come out on top.

Indian water purifiers are sold inside a service culture with no equivalent in China. Every major brand offers an annual maintenance contract, usually quoted in the same breath as the machine, and they come in two shapes: a labour-only contract covers visits, diagnosis and fitting but bills you for every cartridge, while a comprehensive contract folds the scheduled consumables in. Only the comprehensive one can be compared like for like against doing it yourself.

What follows is a model, not a price list — the catalogue carries no service prices. Get two real quotes for your machine in your city and substitute them, because the spread between brands and cities is wider than the spread between levels.

What the AMC premium buys is not filtration — it is somebody arriving with the right cartridge, a tank clean you would not otherwise do, and a single number to budget. For many households that is worth paying for, and the honest reason is not laziness: the failure mode of self-service is not doing it, and not doing it is what kills membranes.

Two traps. An AMC quoted at the point of sale is frequently a first-year promotional price, so ask what year two costs. And "comprehensive" varies: some contracts include the membrane and some schedule it as a chargeable extra, precisely because it is the expensive item. Get that one in writing.

The Chinese half works nothing like this. Those seven rows use composite cartridges that twist out in one motion, bought from the platform that sold the machine and often reordered from the brand’s app — 352’s K10 800G makes a selling point of honest filter-life tracking. No visit, no contract, five minutes of your own labour. The trade is that nobody is coming to check the machine, so nobody will notice a problem you do not notice yourself.

The catalogue carries no wattage field, so this is arithmetic from stated assumptions. Take a booster pump at 50 W — the usual range is a few tens of watts — and a family drawing 15 litres a day from a machine rated at 15 L/h. The pump runs about an hour a day: 0.05 kWh, roughly 18 kWh a year. At ₹8 per unit that is about ₹145 a year in India; at ¥0.55, about ¥10 in China. Against ₹15,999 or ¥2,099 of purchase price, noise.

Add a little for a UV lamp that stays lit and for tanked machines cycling the pump to top up a tank that has bled pressure. None of it changes the conclusion: electricity is the smallest line on an RO bill, and the one row that uses none — the Tata Swach gravity filter — saves an amount that will not pay for itself in a decade.

Hot water is the exception, by physics rather than engineering. Two rows dispense it: the A. O. Smith Z8 Green RO in India and the Xiaomi countertop dispenser in China. Heating ten cups — 2.5 litres — from 25 °C to 95 °C takes about 732 kJ, or 0.2 kWh a day and roughly 74 kWh a year: four times the pump, about ₹595 a year at ₹8 per unit or ¥41 at ¥0.55.

The same pattern drives air purifier running costs : the filter dominates, the motor is the small line, and the thing people worry about matters least.

RO works by pushing water through a membrane hard enough to beat its own osmotic pressure. The water goes through; the dissolved salts do not, and leave instead as a continuous stream of concentrate down the drain. The proportion arriving on the clean side is the recovery ratio, and it decides how much water your drinking water costs.

The catalogue carries no recovery figure for any row, so treat these as ranges rather than specifications. Older and cheaper domestic designs typically land around 20–30 per cent recovery — roughly three litres rejected per litre delivered. Better-engineered machines, with higher pump pressure and a calibrated flow restrictor, are usually quoted at 40–60 per cent, about one for one. Where a given machine sits inside those bands depends heavily on input TDS. The difference is not marginal: a household drawing 20 litres a day rejects some 21,900 litres a year at 25 per cent recovery against some 7,300 at 50 per cent — two machines that look identical on a shelf, roughly 14,000 litres of mains water apart.

Recovery is not simply set as high as the designer likes. The harder you push it, the more concentrated the reject stream becomes and the closer it comes to scaling the membrane, so on a high-TDS borewell supply a machine that rejects generously is protecting an expensive element rather than being wasteful. In money the stream is usually small and occasionally not: Indian tariffs are low and much domestic supply unmetered, so 21,900 litres may cost nothing measurable — until you are buying tanker water in summer, at which point it becomes the largest running cost you have. In China, residential water at roughly ¥4–5 per cubic metre puts the same 21.9 m³ at about ¥90–110 a year.

Seven of the thirteen rows carry the catalogue’s eco flag, which is an either/or: low waste-water ratio **or** non-electric. Six are RO machines earning it on the ratio — Pureit Copper+ Eco and the A. O. Smith Z8 Green in India, and the Xiaomi Mi 1000G, 352 K10 800G, Midea Baize 800G and Angel A7 Lite Master in China. The seventh, the Tata Swach gravity filter, qualifies on the other branch: no electricity, and no reject stream at all because it has no membrane. Do not read the flag as "seven machines with a low waste ratio".

The two Indian eco rows claim specifics — Pureit’s Copper+ Eco saves 60 per cent more water than typical RO, the Z8 Green claims near-zero wastage. Both are manufacturer claims under manufacturer conditions, and "near-zero wastage" usually means the reject is diverted to a tank for washing rather than eliminated: better, but a plumbing decision rather than a suspension of osmosis.

Purchase prices below are exact catalogue figures. The upkeep column takes the model above at the middle of each band, assuming you buy cartridges yourself: ₹1,500 a year for Indian level 1, ₹2,800 for level 2, ₹4,500 for level 3; ¥250, ¥400 and ¥750 for the Chinese equivalents. The Tata Swach line assumes two gravity cartridges a year at around ₹400.

Three things fall out of that table. First, upkeep is around a third of the three-year total for every RO row, in both markets, at every level. That consistency is partly an artefact of bands built to scale with complexity, but the order of magnitude is not — and if a salesperson implies upkeep is a minor consideration, this is the arithmetic to hold against it.

Second, the gravity row breaks the proportion: at ₹2,449 the Tata Swach costs ₹2,400 to keep running for three years — nearly as much again as the machine, about half the total. It is still by far the cheapest line here; it simply is not a seventh of the Kent the way the sticker suggests.

Third, gaps widen. The Z8 Green sits ₹9,401 above the Kent Grand Plus at purchase; over three years the gap grows to ₹14,501, because the level-3 machine spends more every year to stay level. That does not make it a bad buy — hot water, mineraliser, one of the two eco-flagged Indian rows — but the comparison at the counter understates the decision.

Add a comprehensive AMC and each Indian row moves up by roughly ₹3,000 to ₹9,000 over three years. Miss a pre-filter change and replace a membrane early and you erase the entire difference between the level bands in one visit. The cheapest machine to run is the one whose cartridges you actually change.

How to choose a water purifier puts this alongside the other decisions, and storage tanks and flow rates covers the structural difference between the Indian and Chinese halves of this list.