RO Recovery Rate: Balancing Yield and Scaling

Recovery is limited by the solubility of the salts left behind. As water leaves as permeate, calcium, sulphate, silica, barium and strontium concentrate in the brine. Calcium carbonate scaling is predicted by the Langelier saturation index, calcium sulphate by its own saturation index, and silica by its solubility, which falls as temperature drops. Membrane warranties set maximum concentration factors, and those limits, not the pump, usually decide the design.

Small single-element systems typically run at 50–60%, while larger multi-element arrays reach 70–75%. Beyond that, the concentrate becomes aggressive: at 80% recovery the concentration factor is 5, and a feed with 500 mg/L TDS leaves a brine at 2,500 mg/L. Every further point of recovery buys less wastewater but shortens the interval between cleanings.

The concentration factor is 1 divided by (1 − recovery). It assumes the membrane passes water but no salt, which is close enough for design screening. The table shows the effect on a feed of 500 mg/L TDS. Note that the jump from 65% to 75% raises the brine from 2.9 to 4.0 times the feed, a much larger step than the ten-point recovery change suggests.

Feed temperature changes flux by about 3% per °C. A system rated at 25 °C and 1,000 L/h delivers roughly 700 L/h at 15 °C and 1,150 L/h at 30 °C at the same pressure. Cold water also raises viscosity, which increases the pressure needed and can push the pump to its limit before the design recovery is reached. Silica solubility falls with temperature, so a 75% recovery that works at 25 °C may exceed the silica limit at 10 °C.

This is why recovery should be quoted with a temperature. A skid sold as 75% recovery at 25 °C is not a 75% skid in a cold northern plant room, and the concentrate flow will be higher than the datasheet suggests. Design at the coldest month, not at the annual average.

Higher recovery is a pretreatment decision before it is a pump decision. Each limiting salt has its own response, and the design must target the salt that saturates first, not the one that is easiest to remove. A softener removes calcium and magnesium and therefore lifts the carbonate and sulphate limits; antiscalant extends the same limits without removing the ions; pH adjustment shifts carbonate and silica equilibria in opposite directions.

Treat the limiting salt first, then change the array. Softening is the most reliable route: removing calcium and magnesium allows 75–85% recovery on many feeds, at the cost of salt and a brine discharge of its own. A two-stage array is the second route: the concentrate from the first stage becomes the feed of the second, so the same water passes two membrane banks before leaving.

Lower recovery is not a failure. Small systems at 0.5–2 m³/h usually run at 50–60% because a single-element array cannot concentrate further without scaling, and because the capital cost of a two-stage design is hard to justify. High-silica or high-barium feeds often sit at 50–65% by necessity. Sites with cheap water and strict discharge limits may prefer a larger concentrate volume at lower TDS, which is easier to reuse or discharge.

The right question is not what recovery is possible, but what recovery is economical after counting water cost, discharge fees, pretreatment chemicals, cleaning labour and membrane replacement. A 75% system that cleans every month can cost more than a 60% system that cleans twice a year.

Use calibrated flow meters on permeate and concentrate, or on permeate and feed, with an accuracy of about 2%. Calculate recovery from the two readings and compare it with the design value at the same temperature and pressure. If the feed meter is missing, permeate divided by permeate plus concentrate gives the same result without assuming a feed flow.

Record recovery daily together with pressure, temperature and conductivity. A fall in recovery at constant pressure usually means fouling or a partially closed valve; a rise in conductivity at constant recovery means salt passage has increased, which points to membrane damage or oxidation. Neither trend is visible from a single weekly reading.