The Purifier Lab · Water
How much does RO rejection actually fall as a membrane ages?
Over twelve months on a real Indian domestic supply, how does a standard RO membrane's rejection rate change — and at what point does treated-water TDS cross the level the household bought the system to avoid?
Results
No results yet. This page publishes the protocol only.
We have written and published the method below, and we have not yet run it. There are no measurements on this page, no charts and no percentage claims, because there is nothing to report.
The protocol is published first deliberately. It means we cannot quietly amend the method once we have seen a result we did not like, and it lets anyone — including manufacturers — tell us the method is wrong before we run it. If you can see a flaw in it, tell us.
The protocol
Log feed, product and reject TDS weekly on a domestic RO system for twelve months, computing rejection percentage at each point, and record every service event including filter and membrane changes.
Instrument
Calibrated handheld TDS meter with automatic temperature compensation
Model: To be stated on first published run
Calibration: Checked against a 342 ppm NaCl standard solution monthly. Any drift beyond 5% triggers recalibration and the affected weeks are flagged in the published data.
What is held constant
- Same meter throughout the whole period. Two meters with different conversion factors would make the series meaningless.
- Same sampling points every week: the feed line before the sediment filter, the product tap, and the reject line.
- Same time of day, because supply pressure and feed TDS both vary through the day on intermittent municipal supply.
- Product sample drawn after running the tap for 30 seconds, so the reading is fresh permeate rather than water that has sat in the tank.
- Water temperature recorded at every reading. Conductivity rises roughly 2% per °C and an uncorrected series would show a false seasonal trend.
- Every service event logged with date and part — sediment, carbon, membrane, post-carbon, UV lamp.
Steps
- 1
Establish the baseline
Log feed, product and reject TDS on a newly installed or newly re-membraned system, at three readings across a week, to establish a starting rejection rate rather than a single point.
- 2
Weekly logging
Same day, same time, same taps, same 30-second flush. Record temperature with every reading. Missing a week is recorded as missing rather than interpolated.
- 3
Log every service event
Filter changes reset part of the system, and a rejection change immediately after a carbon change is a different fact from one that developed over eight weeks. Undated service records make the whole series uninterpretable.
- 4
Cross-check quarterly
Send a product-water sample to a laboratory for gravimetric TDS every three months, so the handheld meter series is anchored to a reference method at four points.
- 5
Publish the full series
Every weekly reading, temperature, computed rejection, service events, and the quarterly laboratory cross-checks — including any week where the meter and the laboratory disagreed.
What this test cannot tell you
Published with the protocol rather than buried after the results. A test whose limits are not stated is a claim, not a measurement.
- One system on one supply. Membrane life depends heavily on feed water quality, and a result from municipal supply says little about a high-iron borewell.
- A handheld meter infers TDS from conductivity. It is adequate for tracking change on the same water with the same meter, and it is not a substitute for gravimetric measurement — which is why the quarterly laboratory cross-check exists.
- Rejection percentage depends on feed pressure, which is not controlled here. A drop in municipal pressure will show up as a drop in rejection that is not the membrane's fault.
- TDS rejection is not the same as rejection of any specific contaminant. A membrane still rejecting 90% of dissolved salts may have lost more of its rejection of a particular ion, and this protocol cannot see that.
- Twelve months is one membrane's story. It is a case study, not a life expectancy for the model.
Note