Gel Fines at the RO Skid: What a Fouling Autopsy Actually Shows

In March a 220 m3/day plant pulled its first-stage RO elements after eleven weeks instead of the usual nine months. The lead element came out with a beige, jelly-like film, and the feed pressure had climbed from 1.8 to 4.6 bar. Nothing in the antiscalant programme had changed. What changed was upstream: three weeks of solidification trials on the equalisation tank, drawing from the same sump that feeds the RO.

That is the failure worth understanding properly. Not “polymer ruins membranes” as a slogan, but the specific path fines take from a mixing tank to a spiral-wound element, and the four places where a plant can intercept them without buying anything new.

Where the fines actually come from

Dry superabsorbent powder is a distribution, not a single particle size. A typical industrial grade runs from about 50 to 400 microns, and the fine tail is the part that misbehaves. When powder is drawn into a venturi eductor too fast, or dropped into a tank with a slow-turning paddle, a fraction of the grains never see free water long enough to wet through. They form skins instead: a dry core inside a swollen shell. Those particles stay soft, stay small, and stay suspended far longer than a properly hydrated gel.

Mechanical shear does the rest. A centrifugal transfer pump running at 2,900 rpm chops partly hydrated gel into fragments that pass a 200 micron screen without effort. So the plant ends up with a stream that looks clear, tests normal for suspended solids on a gravimetric filter, and still carries several hundred milligrams per litre of soft organic gel. Standard SDI measurements under-read it because the gel deforms and squeezes through the filter paper.

The prefilter is not the safety net people assume

Most leachate RO trains have a 5 micron cartridge guard before the high-pressure pump, and operators reasonably assume it stops anything of consequence. Two things go wrong. First, cartridge ratings are usually nominal, not absolute, so a 5 micron nominal element passes a meaningful share of 5 micron particles. Second, and worse, a loaded cartridge does not fail politely. It compresses, channels, and lets a concentrated slug through at the exact moment the differential pressure alarm is already ignored because it has been at 0.9 bar for a fortnight.

There is a second failure pattern that catches plants out. Gel that accumulates on the cartridge face is compressible. Under 0.5 bar it forms a tight mat that blinds the element in days; under a higher pressure drop it extrudes and passes. So the filter either blocks too early and costs you cartridges every shift, or it passes the problem downstream. Neither outcome is visible on a daily log sheet.

What an autopsy can and cannot tell you

Send a fouled element for analysis and you will get a list: loss on ignition, elemental scan, sometimes an FTIR spectrum. High organic LOI with sodium and no significant calcium or silica points toward gel rather than hardness scale or biofilm. That is useful. What the report cannot do is name the day the gel arrived or the dose that caused it, because the deposit integrates weeks of operation.

The distinction matters because biofilm and gel look similar on a photograph and behave nothing alike in cleaning. A biological film responds to caustic and biocide. Cross-linked polyacrylate does not dissolve; it swells further, and a caustic clean can push a marginal element into a permanently higher pressure drop. Plants that treat a gel fouling event as routine CIP often destroy the element they were trying to save.

Four interception points, cheapest first

Start at the point of wetting. A eductor with sufficient motive water, powder fed at a controlled rate, and a full thirty seconds of contact before the slurry reaches a pump will remove most of the fine fraction at source. This costs discipline, not capital.

Next, give the slurry somewhere to settle. A hold tank with a quiescent zone and a bottom drain lets unhydrated or partly hydrated particles drop out before transfer. Third, use a settling or screening step sized for the actual flow rather than the catalogue. Fourth, and only fourth, add a guard filter with an absolute rating and a differential pressure trend that somebody actually reads.

One plant we know cut membrane cleaning frequency from every five weeks to twice a year by adding nothing but a 20-minute settling hold and a 100 micron inline screen ahead of the feed tank. The dose did not change. The placement did.

The next time a batch is destined for a membrane train rather than a truck, log three things on the batch sheet: powder feed rate, contact time before the transfer pump starts, and the clarity of a settled sample at ten minutes. If you are specifying polymer for a plant with an RO downstream, the SAP powder product range lists particle size distribution for each grade, which is the number that decides how much of your dose ends up on an element.