SDI and Turbidity: Setting RO Feed Water Limits

Turbidity measures light scattered by suspended particles, usually reported in NTU. It is dominated by particles around 1 µm and larger, and it is insensitive to the submicron colloids that pack tightly into a membrane spacer. Water at 0.3 NTU can still carry enough colloidal silica or iron to foul an array, while water at 1.2 NTU of coarse silt may be removed easily by a 5 µm cartridge.

Turbidity is still useful: it is continuous, cheap and fast, and it catches upsets such as filter breakthrough. The mistake is treating a low NTU as permission to skip SDI measurement. Turbidity tells you whether the filter is working now; SDI tells you whether the water is safe for the membrane over months.

Turbidity also depends on the instrument. A portable meter and an online probe can differ by 0.1–0.3 NTU on the same sample, which matters when the limit is 1 NTU. Calibrate against a formazin standard and use the same instrument for trend comparison.

The limits below reflect common membrane warranty conditions, not a single universal standard. The enforceable values are in the element datasheet and the project warranty. What matters is that SDI and turbidity are only two of several limits: a design that meets both while missing chlorine or iron still fails in the first year.

Coagulation with a metal salt or polymer, followed by multimedia filtration, typically brings SDI15 from 6–8 down to 3–5. Ultrafiltration, with a nominal pore size of 0.01–0.1 µm, typically delivers SDI15 below 3 and turbidity below 0.1 NTU, and it does so more consistently than media filters when the raw water changes. A 5 µm cartridge after either step is a guard, not a treatment stage: it protects the membranes from a media upset but does not lower SDI by itself.

Start with the test itself. A dirty filter holder, a damaged disc, a leaking O-ring or a pressure that drifts below 2.1 bar will all produce a wrong number, and the operator will then chase a pretreatment problem that does not exist. Clean the holder, use a new disc, confirm the pressure and record the water temperature before changing any dose.

If SDI rises only after a backwash, the problem is the backwash sequence, not the media. Check the backwash flow, duration and the settling time before the filter returns to service.

Measure SDI15 at commissioning and after any change in source, media or pretreatment. In operation, daily on surface water and weekly on stable municipal or well water, plus after every media replacement and after any turbidity upset. For the formal test, take three readings at 5, 10 and 15 minutes and report the 15-minute value; a single reading hides variability. Log temperature with the result, because SDI rises as water warms and flow changes.

UF produces water that usually passes the SDI limit, but it does not replace the measurement. Membranes can break, modules can leak at the O-rings, and a UF that passes 0.1 NTU can still pass colloids during a fibre break. Keep SDI as a periodic check and use continuous turbidity and particle counting as the early warning. The combination is stronger than either test alone.

A UF skid also produces its own waste stream: backwash and chemically enhanced backwash typically reject 5–15% of the feed, and that volume belongs in the water balance before the RO is sized.

Chlorine is the most destructive: polyamide membranes tolerate less than 0.1 mg/L of free chlorine, and the damage is irreversible. Iron and manganese foul the spacer, hardness and silica scale the concentrate end, and temperature sets both flux and solubility. A feed that passes SDI but carries 0.5 mg/L chlorine will fail faster than a feed with SDI 5 and no oxidant.