Silt Density Index (SDI): What the RO Feed Water Test Really Tells You

Every RO membrane datasheet carries the same line: feed SDI less than 5, preferably less than 3. Most plants measure it at commissioning, write the number in the log, and never look at it again — until the membranes foul and the supplier asks for the SDI history that doesn't exist. The silt density index is the single most predictive number for RO fouling, it costs almost nothing to measure, and it is routinely misunderstood. Here is what it actually tells you, and what it doesn't.

What SDI measures

SDI is a test of how quickly a 0.45-micron filter plugs when water is pushed through it at 2.1 bar (30 psi). You time how long it takes to collect 500 mL at the start, wait 15 minutes with the water still flowing, then time 500 mL again. The formula is:

SDI = [1 − (t₁ / t₂)] × 100 / 15

where t₁ is the initial fill time and t₂ the fill time after 15 minutes. If the filter barely plugs, t₂ ≈ t₁ and SDI is near zero. If it plugs badly, t₂ is much longer than t₁ and SDI heads toward its maximum of 6.67. The number is dimensionless and the test is standardised as ASTM D4189.

What SDI is really measuring is the colloidal and fine particulate load in the water — the material between roughly 0.1 and a few microns that is too small to settle, too small to be caught by a sand filter, and exactly the right size to blind an RO membrane's feed channel. That is why it predicts fouling far better than turbidity does.

Why turbidity is not enough

Turbidity measures how much light the water scatters, in NTU. It is dominated by larger particles and it is blind to the fine colloids that matter most to an RO. Water at 0.3 NTU can have an SDI of 6, and a plant that meets its turbidity specification can foul a membrane in weeks. The two numbers answer different questions. Turbidity tells you whether the clarifier and filters are working on the coarse fraction; SDI tells you whether what is left will foul a membrane. A well-run RO measures both, and treats an SDI rise as the earlier warning.

What the numbers mean in practice

• SDI below 1. Exceptional. Typical of RO permeate, or of a well-run UF. Membranes will run years between cleans.
• SDI 1–3. Good. The realistic target for any RO with a proper pre-treatment train. Minimal colloidal fouling; cleaning frequency set by biology or scaling instead.
• SDI 3–5. Acceptable per the datasheet, but the membranes are working. Expect cleans every one to three months and a faster decline in flux.
• SDI above 5. Outside warranty on most elements. Fouling will be rapid and eventually irreversible. Do not run an RO on this feed for long.

Two practical notes. First, SDI is not linear — an SDI of 4 is not "twice as bad" as 2; the difference in fouling rate is considerably more than that. Second, the SDI of the same water can vary hour to hour with clarifier performance, backwash cycles and raw water turbidity. A single reading proves little; a daily reading, trended alongside the normalised RO data, is what makes the number useful.

How pre-treatment moves SDI

SDI is the report card for everything upstream of the RO:

• Clarification and coagulation convert colloids into flocs that settle or filter out. A dosing pump that stops is often visible in the SDI within hours.
• Multimedia / pressure sand filters remove the flocs but do little to raw colloids; on a surface water or treated effluent, a sand filter alone typically delivers an SDI of 3–5 on a good day and worse after a backwash.
• Cartridge filters (5 micron) are a last line of defence, not a treatment step. A rapidly rising cartridge ΔP is a sign the SDI upstream has gone wrong.
• Ultrafiltration is the only conventional pre-treatment that reliably delivers SDI below 2–3 regardless of raw water variation, which is the whole argument for UF ahead of RO on variable or effluent-derived feeds.

How to measure it properly

The test is simple, but it is easy to get wrong. The pressure must be held at exactly 2.1 bar — a regulator on the sample line, not a guess. The sample point should be immediately before the RO, after the cartridge filters, with the line flushed first. The water temperature should be noted, because viscosity changes the fill time. If the initial 500 mL takes more than a minute, the water is too dirty for a meaningful 15-minute test; use a 10- or 5-minute interval and record which one you used. And the filter itself matters — use the specified 0.45-micron, 47 mm membrane discs, not whatever is in the lab drawer. A plant that measures SDI daily at the same point, same time, same method, has an early-warning system for its membranes that costs a few rupees per test.

SDI is not a compliance number; it is a diagnostic. It tells you, before the RO does, whether the pre-treatment is keeping its promise. If your SDI is drifting above 3, or you are fouling membranes with an SDI you are told is fine, that is a conversation worth having with us — [email protected] or +91-98100 00233.