Quick answer: Reverse osmosis and EDI are not competing technologies — they occupy different positions in the same treatment train. RO removes the bulk of dissolved solids and is the workhorse of every industrial water system. EDI replaces the mixed-bed ion exchange that used to follow it, producing ultrapure water continuously without the chemical regeneration cycle. Choosing between them is really a question of how pure the water must be, and whether you can accept a chemical regeneration workflow.
What each technology actually removes
Both processes remove dissolved ions, but through different mechanisms with different limits, and understanding that difference explains nearly every selection decision.
Reverse osmosis applies pressure to force water through a semi-permeable membrane while rejecting dissolved salts, colloids and most organic molecules. It removes the large majority of dissolved solids in a single pass, and it is the only one of the two that handles the bulk of the ionic load.
Ion exchange swaps ions in solution for ions on a resin bed. Traditional mixed-bed deionisation produces extremely high-purity water, but the resin saturates and must be regenerated with acid and caustic — a chemical-handling process that produces a concentrated waste stream.
Electrodeionisation (EDI) combines ion exchange resin with an electric field. The applied current continuously drives captured ions out of the resin and into a concentrate stream, so the resin regenerates in place while the unit operates. There is no acid or caustic regeneration cycle, and the process runs continuously.
The constraint that governs EDI feasibility
EDI is not a standalone process. It requires feed water that already meets specific quality limits — typically produced by a preceding RO stage — and its performance depends on that feed quality being maintained. This is the single most important design fact about EDI: it is a polishing stage, and it fails if its feed water is out of specification.
| Parameter | Why it matters to EDI | Consequence if exceeded |
|---|---|---|
| Feed conductivity | Determines the ionic load the EDI stack must carry | Product quality drifts above target; stack may not reach rated resistivity |
| Hardness (calcium, magnesium) | Can precipitate at the high pH near the cathode | Scale forms inside the stack and permanently degrades it |
| Silica | Poorly removed, accumulates in the concentrate path | Progressive loss of performance, hard to reverse |
| CO2 and TOC | Increase the ionic load and foul resin | Reduced resistivity; organic fouling of the resin bed |
| Free chlorine and oxidants | Attack the ion exchange resin irreversibly | Permanent stack destruction |
Hardness scaling is the failure mode that matters most in practice, because it is cumulative and irreversible. If the RO stage upstream of an EDI unit is allowed to operate with membrane elements that have lost their rejection performance, or if softening is omitted where the raw water hardness is high, the EDI stack will scale. A scaled stack is not recovered by cleaning — it is replaced. This makes the RO and any softening stage genuinely critical to the EDI's service life, not merely upstream equipment.
How to choose: the decision framework
For most industrial projects the selection follows a fairly short path once the required product quality and the feed water analysis are known.
| Required quality | Typical target | Recommended train | Notes |
|---|---|---|---|
| Process water, general industrial | Conductivity in the tens of µS/cm | Pretreatment → RO | Simplest and lowest operating cost; adequate for many production uses |
| Boiler feed water (medium pressure) | Low hardness and conductivity | Softening → RO | Softening protects against hardness carryover; RO handles dissolved solids |
| High-purity process water | Resistivity above 10 MΩ·cm | Pretreatment → RO → EDI | Continuous operation, no chemical regeneration |
| Ultrapure water, pharmaceutical or electronics | 18 MΩ·cm with TOC and microbial control | Multi-stage pretreatment → double-pass RO → EDI → polishing (UV, ultrafiltration) | Requires continuous monitoring and validated operation |
| Existing mixed-bed DI plant | Same as current output | RO → EDI replacing the DI stage | Eliminates chemical regeneration; requires feed water within EDI limits |
The most common specification error is treating EDI as a substitute for adequate RO. It is not. EDI's product quality target can only be reached if its feed stays within its design envelope, so specifying an EDI stack without first confirming that the upstream RO will reliably hold feed conductivity, hardness and silica within limits sets the project up for a replacement cost months into operation.
The second most common error is the reverse: overspecifying. If the process genuinely only needs water in the tens of µS/cm range, an RO-only train with appropriate pretreatment will meet it at lower capital and operating cost, and with less to maintain. Adding an EDI stage "for margin" adds a component whose feed-water tolerances must then be managed for the life of the plant.
Operating and maintenance differences that matter
The operating profile of the two stages differs enough that maintenance planning should treat them separately.
- RO maintenance centres on pretreatment performance, antiscalant dosing accuracy, membrane cleaning cycles and element replacement. Membrane elements typically last several years when feed conditions are held within design limits, and foul more quickly when they are not.
- EDI maintenance is comparatively low-touch — no resin regeneration, no chemical storage — but it depends absolutely on feed quality. Monitoring feed conductivity, hardness and TOC is the maintenance activity, because the stack cannot be cleaned once scaled.
- Instrumentation is not optional. Conductivity, flow, pressure differential and temperature monitoring on the RO stage, plus product resistivity on the EDI stage, are what allow a drift to be caught before it becomes damage.
- Consumables differ. RO consumes antiscalant, cleaning chemicals and eventually membrane elements. EDI consumes almost nothing in routine operation, which is precisely why it is preferred where chemical handling is undesirable.
What to put in the specification
To receive a train that performs as intended, the specification should state the following:
- Product water quality with the measurement basis — conductivity or resistivity, and at what temperature it is stated
- Feed water analysis in full — including hardness, silica, iron, CO2, TOC and oxidant residual, since these govern EDI feasibility
- Required flow, recovery rate and reject handling — recovery affects feed quality through concentration, so it belongs in the specification, not in the supplier's discretion
- Whether chemical regeneration is acceptable on site — this single question decides whether EDI is appropriate, independent of water quality targets
- Monitoring and control scope — which parameters are continuously measured, logged and alarmed, and whether the system reports to a plant control system
- Service access and consumable supply — who cleans membranes, who supplies elements, and what the expected replacement interval is under the specified feed conditions
Specifying an RO or RO+EDI train? Mingtao engineers can review your feed-water analysis and propose a process train with a clear statement of what each stage is required to achieve — so the EDI stage, if it is included, has feed water it can actually tolerate. Send your water analysis and target quality for a no-obligation proposal.
Frequently asked questions
Do I need EDI if RO already meets my conductivity target?
No. If RO-only product water meets your process requirement, adding EDI gives you no benefit and adds a component with its own feed-water tolerances to manage. EDI earns its place when you need resistivity in the high-megohm range, or when you want to eliminate the chemical regeneration cycle that mixed-bed deionisation requires — not as a general upgrade.
Can EDI replace an existing mixed-bed deionisation plant?
Often yes, as a retrofit downstream of existing RO. The critical check is whether the existing RO reliably delivers feed water within the EDI unit's specification — particularly for hardness, silica and oxidant residual. If the existing RO does not, the retrofit needs to address that first, because an EDI stack scaled by hardness carryover cannot be recovered.
Why does EDI have a CO2 limit?
Dissolved carbon dioxide passes through RO membranes and dissolves to form bicarbonate and carbonate ions, which add to the ionic load the EDI stack must remove. High CO2 therefore raises the current requirement and can prevent the stack from reaching its rated resistivity. Where CO2 is high, a degasser between the RO and EDI stages is a standard remedy.
What is the typical service life of RO membranes and EDI stacks?
Both depend far more on feed-water control than on time. RO elements under well-managed pretreatment commonly serve several years before replacement; the same elements under fouling conditions may need replacement in a fraction of that. EDI stacks under correct feed conditions last many years, but are destroyed quickly by hardness scaling or oxidant exposure. This is why monitoring instrumentation and pretreatment discipline matter more than the nominal ratings of either component.
Is EDI more expensive to operate than mixed-bed deionisation?
The comparison depends on site conditions. EDI consumes electrical power for the stack and slightly more RO feed water for its concentrate stream, but it eliminates acid and caustic purchase, storage, handling, regeneration waste neutralisation and the labour associated with regeneration. In facilities where chemical handling is costly or restricted, EDI is frequently the lower total operating cost despite a higher capital cost.
