A1 · How to Size an Industrial RO System: The 6 Inputs Every Spec Must Include

By Ella Liu — Industrial Water Treatment Specialist at Rong Water · [email protected]

The 30-second answer

Picture this: you ask three vendors for a 100 m³/h RO system. Two weeks later you get three quotes. None of them will deliver the same water at the same OPEX — even though they all wrote “100 m³/h” on the cover page.

Why? Because the spec didn’t pin the six inputs that drive the whole design — the numbers you have to know before any vendor, including us, can give you a meaningful quote. If a vendor quotes you on a single figure (“we want 100 m³/h”), the quote you receive is a guess.

Here they are:

1. Feed water source and analysis (TDS, hardness, iron, silica, free chlorine, SDI, temperature)
2. Required permeate flow (m³/h or m³/d)
3. Required permeate quality (conductivity / resistivity, plus any specific ion limits like silica or TOC)
4. Recovery rate target (typically 65–85 % for brackish water, 35–50 % for seawater)
5. Feed temperature range (winter minimum drives membrane count; summer maximum drives pump head)
6. Energy recovery requirement (only relevant for seawater)

Every other parameter on a vendor’s quote sheet — number of vessels, pressure rating, pump power, OPEX — falls out of these six. This article walks through each, with the arithmetic on a 100 m³/h brackish water (BWRO) worked example so you can follow the same logic on your own plant.

1. Feed water source and analysis — the input that breaks most projects

A vendor quoted me once on a first call — no feed analysis, just “assume city water”. That assumption is exactly where the project went sideways later. Feed water analysis is the input that breaks most projects, because vendors usually work from a generic TDS figure rather than a full ion-by-ion breakdown. The minimum set of parameters you must supply:

Two non-obvious traps that cost projects:

• Free chlorine > 0.1 mg/L will destroy a polyamide membrane in weeks. Even a municipal tap treated with chlorine will damage an unprotected RO. Always specify either dechlorination (activated carbon, sodium bisulfite) or a chlorine-tolerant membrane — but the latter is rare in industrial use. For pharmaceutical and electronics buyers, dechlorination is non-negotiable.
• SDI is more important than TDS. A well water with TDS 1,500 mg/L and SDI 5 will foul faster than seawater with TDS 35,000 mg/L and SDI 2. If you can only test two things, test TDS and SDI. For buyers sourcing from surface water or food & beverage plants where organic load is variable, SDI control through UF pre-treatment (UF-2000) is the standard fix.

For our 100 m³/h worked example we will assume a brackish well water at TDS 2,500 mg/L, hardness 200 mg/L as CaCO₃, iron 0.05 mg/L, silica 15 mg/L, no chlorine, SDI 2.5, temperature 20 °C winter minimum. This is a typical industrial-user feed in inland China, and close enough to a Middle East brackish aquifer profile to be representative.

2. Required permeate flow — the question vendors ask first

Permeate flow is the easiest number to specify and the most often misunderstood. The classic mix-up: you write “100 m³/h” but you actually mean “100 m³ over a 10-hour shift”. Those are different by a factor of ten — the first is a continuous-duty plant, the second a batch plant running one design flow a fraction of the time.

Two ways people state it:

• Permeate flow in m³/h (instantaneous) — for continuous-duty plants.
• Permeate flow in m³/d — for batch or shift-duty plants; convert to m³/h by dividing by the planned daily operating hours.

For this example: 100 m³/h continuous, 24 h/d → 2,400 m³/d permeate.

3. Required permeate quality — conductivity is the floor, not the spec

For industrial applications, “RO permeate quality” is not one number — it is a list of limits, and the most stringent one drives the design. Typical targets:

For brackish water at TDS 2,500 mg/L, a single-pass RO gives 5–30 μS/cm — comfortably meeting boiler feed and most process needs, but not semiconductor or pharma-grade. To reach < 1 μS/cm you need either two-pass RO, or single-pass RO followed by EDI.

For boiler-feed applications where silica and hardness scaling are concerns (see boiler feedwater and power industries), a single-pass RO followed by a polishing mixed bed or EDI is the standard. The deionized-water train we ship for this is the DI-1000 series.

For this example: 100 m³/h permeate at < 10 μS/cm → single-pass RO, two-stage 2:1 arrangement (the standard brackish configuration). If you instead needed < 1 μS/cm, the design flips to two-pass RO or RO+EDI, and the element count roughly doubles — that is a separate decision tree.

4. Recovery rate — the lever that decides membrane area and concentrate volume

Recovery rate Y is the fraction of feed water that becomes permeate:

Y = Qp / Qf
where Qp = permeate flow, Qf = feed flow

The remainder (1 − Y) is concentrate, and it has to go somewhere. Higher recovery means less concentrate to discharge, but also higher salinity at the tail element, higher scaling risk, and higher osmotic pressure at the back end — so the pump has to push harder.

Typical brackish-water RO targets 70–85 % recovery. Above 85 %, silica and calcium-sulfate scaling limits start biting and you need either better antiscalant chemistry or a concentrate recycle stream.

For a 100 m³/h permeate at 75 % recovery:

Qp = 100 m³/h
Y  = 0.75
Qf = Qp / Y = 100 / 0.75 = 133.3 m³/h
Qb = Qf − Qp = 33.3 m³/h   (concentrate / blowdown)

So a 75 %-recovery system needs a 133 m³/h feed pump and disposes of 33 m³/h concentrate.

5. Feed temperature — the silent design driver

RO membrane flux is strongly temperature-dependent. The standard industry correction is to apply a temperature correction factor (TCF) — at lower temperature, the membrane produces less water per unit area, so you need more membrane area (more elements) to hit the same permeate flow.

Here’s the rough number to carry in your head: at a 25 °C design, every 1 °C drop in feed temperature needs about 0.5–0.7 % more membrane area. Over a 20 °C swing (winter to summer in northern China or central Europe), that’s 10–14 % more area — which shows up as more pressure vessels and higher capex.

Design to the winter minimum, not the annual average. A vendor who designs at 25 °C and quotes you a “100 m³/h system” is really quoting a summer system that will under-produce every winter. That gap is the first thing your operator complains about in January.

For this example, designing at 20 °C means we apply a TCF of about 1.13 versus the 25 °C reference. Element count goes up ~13 % to compensate.

6. Energy recovery — only for seawater

For brackish water, the pump power is modest (high-pressure pump at 1.0–1.5 MPa, 75–80 % efficient). Energy recovery devices (ERDs) — pressure exchangers, turbochargers, Pelton wheels — only pay back when the feed osmotic pressure is high enough that the concentrate stream still carries substantial pressure energy. That is seawater (SWRO), not brackish.

For BWRO at TDS 2,500 mg/L, ERDs are not economic. Skip this input.

For SWRO at TDS 35,000 mg/L, ERDs cut energy consumption by 30–60 % and are essentially mandatory. The two commercial families — isobaric pressure exchangers (PX, Energy Recovery Inc.) and turbochargers / Pelton wheels — have different OPEX, footprint, and tolerance to feed-salinity swings. That choice is a separate decision and will be covered in the SWRO post (Series A5).

The 100 m³/h brackish water worked example — assembled

The skid our team ships most often for this envelope is the RO-4000 series in single-pass, two-stage configuration — 100 m³/h is exactly its design point. For higher-purity output (semiconductor or pharma), the same 100 m³/h permeate feeds into an EDI polishing stage, reaching > 15 MΩ·cm without chemicals. For seawater duty, see seawater reverse osmosis (SWRO) (Series A5, forthcoming).

The 3 mistakes that shorten RO membrane life

1. Designing at summer temperature. Winter derating is not optional. If you do not specify the winter minimum, the system will under-produce cold-season for the first year and you will be retrofitting vessels at year two.
2. Ignoring iron and chlorine. Iron > 0.1 mg/L fouls the front end irreversibly. Free chlorine > 0.1 mg/L destroys the membrane in weeks. Both need explicit treatment upstream — usually a greensand filter for iron and either activated-carbon or sodium-bisulfite dosing for chlorine.
3. Pushing recovery above 80 % on hard feed. At 80 %+ recovery on a 2,500 mg/L brackish feed, calcium-carbonate and silica scaling can take the membrane down by month six if antiscalant dosing is not tuned to the concentrate chemistry.

The 3 numbers to write into your procurement spec

If you remember nothing else from this article, write these three into your RFQ so every vendor is bidding on the same basis:

1. Feed water analysis, ion-by-ion (TDS, hardness, iron, silica, free chlorine, SDI, temperature range).
2. Permeate flow AND permeate quality (μS/cm target, not just “RO water”).
3. Recovery target (the vendor’s default of 75 % is fine if it matches yours — but specify it).

A vendor who refuses to design to a stated recovery target, or who quotes on a single “we want X m³/h” input, is guessing. A correctly-sized RO is calculable. An incorrectly-sized one becomes a maintenance contract you didn’t ask for.

→ Send us your six inputs and we will return a sized RO spec + OPEX within 48 hours.

About the author: Ella Liu — Industrial Water Treatment Specialist at Rong Water (a brand of Shenzhen Qiukanghong Technology Co., Ltd., est. 2012). She has sized and commissioned brackish, seawater and high-purity RO systems for buyers across pharma, electronics, food & beverage, power and desalination. Connect on LinkedIn or email [email protected].