Membrane fouling in RO: types, signs and how to tell them apart

Every reverse osmosis plant fouls. The question is never whether the membranes foul, but what is fouling them, where in the train, and how fast. Get those three answers right and the cleaning regime, the pretreatment and the chemical programme almost choose themselves. Get them wrong and a plant can spend years cleaning with the wrong chemistry while membranes age early.

This article sets out the four main fouling types, the signs each one leaves in your operating data, and a practical way to tell them apart before anyone opens a pressure vessel.

The three numbers that matter

Fouling shows up in three measurements, and almost every diagnosis comes from how they move together:

• Normalised permeate flow — how much water the membranes pass at standard conditions. Falls when the membrane surface is covered.
• Normalised salt passage (or rejection) — how much salt gets through. Rises when scale or damage affects the membrane.
• Differential pressure (ΔP) across each stage — rises when the feed channels are blocked.

The word that matters is normalised. Raw permeate flow changes with temperature, feed salinity and pressure every day; without correcting for those, a cold week looks like fouling and a warm week hides it. If your plant does not normalise its data yet, start there — see RO data normalization explained.

The four main types of fouling

Type	Typical location	Normalised flow	Salt passage	ΔP
Biofouling	First stage, lead elements	Falls	Little change, later rises	Rises, often sharply
Colloidal / particulate	First stage, lead elements	Falls	Little change	Rises
Mineral scaling	Last stage, tail elements	Falls	Rises	Rises slowly
Organic fouling	First stage	Falls	Little change, can fall	Modest rise

These are typical patterns, not rules. Real plants often have two mechanisms at once, which is exactly why the data needs looking at stage by stage.

Biofouling

Bacteria attach to the membrane and spacer, then grow a slimy biofilm. Because it grows, biofouling tends to accelerate: a ΔP trend that bends upwards rather than rising in a straight line is a classic sign. It concentrates in the first stage, where nutrients are highest.

Common contributors: warm feed water, dosed nutrients (some antiscalants and phosphate-based chemicals), intermittent operation and dead legs.

Colloidal and particulate fouling

Fine silt, clay, iron and aluminium flocs or carry-over from coagulation settle on the lead elements. ΔP across the first stage rises and flow falls; rejection is largely unaffected. A high silt density index (SDI) or rising turbidity in the RO feed, or cartridge filters that block quickly, point here. Upsets in the pretreatment usually show up days before the RO responds.

Mineral scaling

Dissolved salts — calcium carbonate, calcium sulphate, barium and strontium sulphate, silica, calcium phosphate — exceed their solubility as the brine concentrates, and precipitate on the membrane. Scaling happens where the water is most concentrated: the last stage, tail elements. Salt passage rises because scale damages the concentration profile at the membrane surface.

Common contributors: recovery set higher than the chemistry allows, antiscalant under-dosing or a dosing pump fault, a change in feed water, or pH drift.

Organic fouling

Natural organic matter (humic and fulvic acids), oil and some polymers adsorb onto the membrane. Flow falls in the first stage with a modest ΔP rise. Over-dosing cationic polymer coagulant in pretreatment is a common, avoidable cause.

A practical diagnosis routine

1. Normalise the data and trend flow, salt passage and ΔP for each stage, not just the whole train.
2. Find where it starts. First-stage problems point to biological, colloidal or organic fouling; last-stage problems point to scaling.
3. Look at the shape of the trend. Steady decline suggests particulates or organics; accelerating ΔP suggests biofouling; a step change suggests an event, such as a pretreatment upset or a dosing failure.
4. Check the pretreatment record for the same period: SDI, turbidity, chlorine or ORP, coagulant dose, antiscalant dose and cartridge filter ΔP.
5. Check the last clean. If a high-pH clean recovered most of the loss, the foulant was probably organic or biological; if a low-pH clean did, inorganic scale or metal oxides are likely. See Is your RO clean working?
6. Confirm with evidence when it matters. Before a membrane replacement or a change of chemical supplier, a membrane autopsy, feed water analysis and a scaling calculation remove the guesswork.

When to act

Membrane suppliers commonly recommend cleaning when normalised permeate flow has fallen by around 10–15%, salt passage has risen by a similar margin, or ΔP has risen by about 15% from the clean baseline. Check the guidance for your own elements. Waiting much longer makes the foulant harder to remove, and each partial clean leaves a little more behind.

The cheapest fouling is the one you notice early. A trend that is reviewed weekly, stage by stage, usually tells you what is happening weeks before a production target is missed.

Where software helps

The routine above is straightforward but laborious by hand, which is why it often does not happen until there is a problem. RO Insight™ applies it continuously: it normalises the data, tracks each stage, separates the likely fouling mechanisms and forecasts when the next clean will be needed. An engineer still makes the call, with the evidence in front of them.