EDI Water Treatment System — Ultrapure Water Without Chemical Regeneration

EDI must be fed with RO permeate, not raw or softened water. Typical limits are conductivity of 10–20 µS/cm or lower, carbon dioxide below about 5 ppm, hardness below 1 ppm as CaCO₃, silica below 0.5 ppm, free chlorine below 0.05 ppm and TOC below 0.5 ppm. Turbidity and SDI should be low enough to protect the resin and membranes. These are screening values: the stack datasheet and warranty conditions take precedence.

Carbon dioxide is the most common reason an EDI underperforms. RO does not reject CO₂ gas, so it passes into the EDI and consumes current without contributing to resistivity. If the feed alkalinity is high, add a degasser, a membrane contactor or pH adjustment before the stack. Hardness and silica are the second common cause: they scale the concentrate side, raise pressure drop and permanently damage the stack.

Both processes can produce ultrapure water. A mixed bed uses a single vessel of cation and anion resin and is regenerated with acid and caustic, either in place or off site. It tolerates a wider feed quality and often has a lower capital cost for small or intermittent duty. EDI regenerates electrically while running, so it avoids chemical handling, neutralisation, regeneration waste and the downtime of a regeneration cycle.

The choice usually comes down to feed reliability and operating pattern. EDI is favoured for continuous, automated duty with stable RO feed; a mixed bed is favoured where feed quality varies, where the plant runs intermittently, or where very low TOC is required and a polishing mixed bed is used after EDI.

The usual train is pretreatment, RO, CO₂ removal, EDI, polishing and a storage or distribution loop. Softening or antiscalant protects the RO; a 5 µm cartridge filter protects the EDI stack; a degasser or membrane contactor removes CO₂ between the RO and EDI. Instruments monitor feed conductivity, product resistivity, flow, pressure and temperature, and interlock the stack off if feed quality falls outside the set points.

Hydraulics matter as much as water chemistry. The stack needs a stable feed pressure and a concentrate flow within the design range; too little concentrate flow causes scaling, too much wastes water. A recirculation loop keeps flow across the stack during low demand. The product line should be short, with no dead legs, and the storage tank should be vented and protected from airborne contamination.

Often it can, but the existing RO must be able to produce consistent low-conductivity permeate, and the CO₂ load must be handled. A retrofit needs space for the stack, a DC power supply, a concentrate drain, new instruments and revised controls. If the existing mixed bed is in good condition, it can remain as a backup or final polisher while EDI takes over the bulk duty, which reduces chemical use without sacrificing redundancy.

A retrofit is not always justified. If the plant runs only a few hours a day, if the feed quality swings widely, or if the existing mixed bed is already cheap to regenerate, the payback may be long. The decision should compare chemical and waste costs, downtime, labour and stack replacement against the capital cost of the retrofit.