What comes out of a healthy plant
A brackish element rejecting 99% of salt at standard conditions does not deliver 1% of feed TDS in practice — the field number runs higher, because rejection is measured against the average concentration along the membrane, which recovery pushes well above the feed value. Working expectations for a healthy brackish system: permeate TDS around 1 – 4% of feed TDS at moderate recovery with high-rejection elements, the upper half of that with low-energy elements or high recovery. A 1 000 mg/L feed typically yields permeate in the tens of mg/L — excellent by any drinking standard, but not zero. Where near-zero matters (boilers, electronics, laboratory), the answer is a second pass or polishing (mixed-bed DI, EDI) downstream, not disappointment in the first pass.
Two chemical quirks of permeate surprise people. CO₂ passes straight through the membrane, so permeate pH is typically 5 – 6.5 and the water is unbuffered and slightly aggressive — pH correction or remineralisation ahead of copper pipework or drinking supply is normal design, not a defect. And permeate is hungry water: low-TDS, low-alkalinity water pulls minerals from cementitious materials, another reason storage and pipework materials deserve a thought.
Normal drift — the movements that are not faults
- Temperature. Salt passage rises with temperature even as production rises: summer permeate is more plentiful and slightly saltier. Judge the plant on normalised numbers, not raw seasonal ones.
- TDS creep on start-up. During shutdown, salt diffuses across the idle membrane; the first minutes of production run measurably saltier until it flushes through. Plants that cycle frequently should divert or tolerate the first permeate; a permeate flush at shutdown largely prevents it.
- Recovery drift. Anything that quietly raises recovery — a throttled concentrate valve, a permeate demand change — raises average feed-side concentration and with it permeate TDS. Check the flows before blaming the membranes.
- Ageing. Rejection declines gently over an element's life; a slow multi-year drift is the membrane amortising, not an event.
Faults — the movements that are
A step change in permeate conductivity is mechanical until proven otherwise, and two different faults hide under that heading. A failed interconnector o-ring or a cracked permeate tube opens a direct path from feed into permeate: one small leak moves the number faster than any membrane chemistry can, and probing the vessel — sampling conductivity along the permeate tube — locates it. A brine seal fitted backwards or damaged does something different: feed slips around the outside of the element instead of being driven through it, so rejection falls and the vessel never builds its normal pressure drop, with no direct feed-to-permeate path to find. Probing finds the first; a vessel running soft on ΔP points at the second. A steady climb with rising ΔP in the last stage says scaling; chlorine damage announces itself as rejection falling plant-wide while flow increases, the membrane quietly becoming an expensive cartridge filter. Each signature points at a different repair, which is why the weekly log of normalised flow, salt passage and per-stage ΔP is the cheapest instrument on the plant.
Rule of thumb: expect permeate around 1 – 4% of feed TDS, slightly acidic, and seasonally variable — that is a healthy plant. Sudden change means mechanical leak, quality-plus-ΔP change means scale, quality-plus-more-flow means oxidation. Normalise before diagnosing.