Pure water field guide: RO and DI systems for plumbers

What a pure-water system is and where it is needed

A pure-water system takes ordinary potable water and removes almost everything dissolved in it, until what comes out the far end is closer to a chemical reagent than to drinking water. Tap water is safe to drink because it carries calcium, magnesium, sodium, chloride, silica, dissolved carbon dioxide, traces of metals, and organic matter at levels a body tolerates. Those same dissolved solids wreck a lab assay, scar a silicon wafer, scale a boiler, or poison a dialysis patient. So a whole class of work exists to take them back out.

You find these systems wherever the water touches something the minerals would ruin. Clinical and research labs feeding analyzers, glasswashers, and autoclaves. Dialysis units, where the water mixes straight into a patient's blood across a membrane. Pharmaceutical suites compounding drugs. Electronics and semiconductor fabs rinsing wafers. Boiler plants and steam humidifiers that scale shut on hard water. Each one demands a purity grade the tap cannot meet, and each one gets there by some combination of reverse osmosis and deionization.

The two big processes are reverse osmosis, which pushes water through a membrane that rejects most dissolved solids, and deionization, which pulls the last ions out on exchange resin. Most real systems run them in series: RO for the bulk, DI or electrodeionization for the polish. Softening, carbon, and sediment filtration sit ahead of the RO to protect it. That broader point-of-entry treatment, sizing a softener and laying out the filters, is its own subject and gets the full walk-through in the water-treatment field guide. This one picks up where high purity becomes the point.

Why does an application need water purer than the tap?

Because the dissolved solids in tap water are inert to a person and ruinous to the process. The application sets a purity the municipal supply was never built to hit, and the gap between the two is the entire reason the system exists.

Work it case by case. In a lab, dissolved ions and organics throw off the result: a few parts per billion of the wrong metal shifts an ICP-MS reading, and residual carbon feeds a baseline drift on an HPLC. In dialysis, the water crosses the dialyzer membrane into the bloodstream, so any chloramine, aluminum, or bacterial endotoxin in it goes into the patient. In semiconductor work, a single dissolved-solid particle bridging a circuit feature is a dead chip, so the rinse water has to be ultrapure or the yield falls. On a boiler, the hardness and silica that ride in with makeup water bake onto the tubes as scale, the scale insulates the metal, and the metal overheats and fails. Steam humidifiers do the same thing in miniature, crusting their electrodes and cylinders.

The common thread is that the contaminant is dissolved, not suspended, so a sediment filter does nothing for it. You cannot strain out calcium ions. You have to either reject them through a membrane or trade them off on resin. That is why pure-water treatment is a different animal from filtration, and why the cost climbs with every grade of purity you add.

How is water purity measured?

Water purity is measured electrically, as resistivity or its inverse, conductivity. Dissolved ions carry current, so the more ions in the water, the better it conducts and the lower its resistance. Pull the ions out and the water stops conducting. That is the whole basis of the gauge: a resistivity meter is really an ion meter reading the bulk water.

Resistivity is given in megohm-cm, and higher is purer. Theoretically perfect water at 25°C tops out near 18.2 megohm-cm, the number the ultrapure world chases, because at that point the only ions left are the water's own from self-ionization. Conductivity runs the other way, given in microsiemens per cm, where lower is purer and that same ultrapure water sits around 0.055 microsiemens. TDS, total dissolved solids in parts per million, is the plain-language version of the same idea, often estimated from conductivity rather than measured directly. A meter that reads resistivity, conductivity, and TDS is reading one physical property three ways.

Temperature matters more than people expect, because conductivity climbs steeply as water warms. A reading taken at 30°C and a reading at 20°C are not comparable unless both are corrected to a 25°C reference, which is why good inline meters do the compensation automatically. The trap with ultrapure water is the opposite of dirty water: dissolved carbon dioxide from the air pulls the resistivity down fast once water leaves the system, so a sample dipped in a beaker reads far worse than the same water inline. Resistivity tells you about ions. It tells you nothing about bacteria, organics, or particles, so it is one gauge among several, not the only one.

Resistivity

Water's resistance to current, in megohm-cm at 25°C; higher is purer, ~18.2 is ultrapure

Conductivity

The inverse of resistivity, in microsiemens per cm; lower is purer, ~0.055 is ultrapure

TDS

Total dissolved solids in parts per million, the plain-language read on dissolved ion load

Purity grades: RO, DI, and ultrapure

There is no single number for pure water. The grade is set by the application and the standard it answers to, and the same system can be overkill for one use and unusable for another. Talk in grades, not absolutes.

RO water, the output of reverse osmosis alone, removes most dissolved solids and lands a few orders of magnitude better than tap. It is fine for many rinse, boiler-feed, and pretreatment duties, and it is the feed for the polishing stages, but it is not ultrapure on its own. DI or deionized water, RO that has been polished on ion-exchange resin, climbs much higher in resistivity and reaches into the lab and electronics grades. Ultrapure water, often the RO-plus-DI or RO-plus-EDI train with UV and final filtration, approaches the 18.2 megohm-cm ceiling and meets the strictest semiconductor and Type I lab needs.

The lab world uses a Type I, II, III split from ASTM D1193 for reagent water, where Type I is the ultrapure grade near 18 megohm-cm, Type II is an intermediate grade, and Type III is roughly RO-grade for general washing and feed. Pharma answers to the USP compendial grades, Purified Water and Water for Injection, which are written around conductivity and total organic carbon limits rather than a resistivity target. Dialysis answers to the AAMI and ISO 23500 water quality standards, which are chemical and microbiological limits, not a single purity number. Pin the grade to the standard the job actually has to pass, and verify the current edition, because the limits and the structure of these standards get revised.

Pretreatment protects the RO membrane

The RO membrane is the most expensive and most fragile part of the system, and pretreatment exists to keep it alive. Skip it and you replace membranes on a schedule measured in months instead of years. Pretreat it right and the membrane does its job quietly for a long time. This is the step that pays for itself, and it is the step a tight budget cuts first and regrets.

Three things in tap water go after the membrane. Chlorine, the disinfectant the utility adds, chemically attacks a thin-film composite membrane and oxidizes it, so activated carbon, or a chemical reducer such as sodium bisulfite, goes ahead of the RO to strip the chlorine and chloramine out first. Hardness, the calcium and magnesium, precipitates as scale on the membrane surface where the water concentrates, so a softener or an antiscalant feed handles it. Sediment and fine particles foul and abrade the membrane, so a sediment filter, commonly in the 1 to 5 micron range, screens them out ahead of everything.

Order matters. Sediment first to protect the carbon and the softener, carbon to kill the chlorine before it can reach either the resin or the membrane, softening or antiscalant to hold the hardness, then the RO. Sizing the softener, choosing carbon over a reducer, and setting the antiscalant dose is the heart of the point-of-entry water-treatment work, and the water-treatment field guide covers that train in full. The one rule that does not bend: chlorine kills the membrane, so the chlorine comes out before the water ever reaches it.

How does reverse osmosis work?

Reverse osmosis pushes water under pressure through a semipermeable membrane that lets water molecules pass while rejecting most of the dissolved solids. Natural osmosis runs water from low concentration to high across such a membrane. Apply enough pressure on the high-concentration side to drive the flow backward, from concentrated toward dilute, and you get reverse osmosis, which is where the name comes from.

The membrane splits the feed into two streams. The water that makes it through, low in dissolved solids, is the permeate, the product you want. The water left behind, now carrying the rejected solids at higher concentration, is the reject or concentrate, and it goes to drain or to a recovery stage. The fraction of feed that becomes permeate is the recovery, and it is a balance: push recovery too high and the reject side concentrates enough to scale the membrane, run it low and you waste water. A typical thin-film composite membrane rejects roughly 95 to 99 percent of dissolved salts, with the exact figure depending on the membrane, the feed water, the temperature, and the operating pressure.

RO does the heavy lifting. It takes the bulk of the dissolved load out in one stage, cheaply and continuously, which is why it sits at the front of nearly every high-purity system. What it does not do is reach ultrapure on its own. A few percent of the dissolved solids pass through, dissolved gases like carbon dioxide slip the membrane entirely, and that residual is exactly what the polishing stages downstream are there to remove. Think of RO as getting you 99 percent of the way and the polish as closing the last gap that the application actually cares about.

Membrane fouling, cleaning, and life

An RO membrane fails slowly, and you read it on the gauges before you read it on a lab result. Watch three numbers: the permeate flow falling off, the differential pressure across the membrane rising, and the salt passage creeping up so the permeate quality drops. Any of those moving in the wrong direction says the membrane is fouling, scaling, or wearing out.

Fouling and scaling are the common killers, and they trace straight back to pretreatment. Scale is hardness and silica precipitating where the reject concentrates, which is why the softener or antiscalant matters. Biofouling is a slime of bacteria growing on the membrane, which is why feed-side bio control matters. Particulate fouling is sediment that got past the prefilter. The defense is the same in every case: pretreat correctly, hold the recovery where the reject does not over-concentrate, and flush the membrane on a schedule so deposits do not set.

When flushing is not enough, the membrane gets a clean-in-place, a CIP, where a cleaning solution circulates through the elements to dissolve scale or strip biofilm, acid for mineral scale and alkaline for organics and bio. Done on time, a CIP buys back most of the lost performance. Let the fouling set hard and no cleaning recovers it, and the elements get replaced. Membrane life varies with the feed water and how well it was protected, so the honest answer to how long they last is that pretreatment and operating discipline decide it, not a number on a spec sheet.

Deionization: the ion-exchange polish

Deionization removes the dissolved ions the RO let through, by trading them for hydrogen and hydroxide on ion-exchange resin. It is a chemical swap, not a filter. The water flows over beads of resin, and the resin grabs the dissolved ions and releases hydrogen and hydroxide in their place, which combine into more water. The result climbs far higher in resistivity than RO alone, which is why DI is the polish that takes the system from clean to high-purity.

There are two resins doing two jobs. Cation resin exchanges the positively charged ions, the calcium, magnesium, and sodium, for hydrogen. Anion resin exchanges the negatively charged ions, the chloride, sulfate, and silica, for hydroxide. Run them in separate beds and you get good water. Mix them together in one bed and you get better water, because the hydrogen and hydroxide neutralize each other on the spot and the exchange runs further toward completion.

DI is a polish, not a bulk process, and that is the point of putting it after RO. Run raw tap water straight onto DI resin and the resin exhausts almost immediately under the full dissolved load, and you are regenerating or swapping tanks constantly. Let RO take out 95 percent or more of the dissolved solids first, and the DI resin only has to handle the small remainder, so it lasts far longer between regenerations. That division of labor, RO for the bulk and DI for the polish, is why the two run in series instead of either one alone.

Mixed-bed and service-exchange DI

Mixed-bed DI is the highest-purity ion-exchange polish, with cation and anion resin blended in one vessel so the exchange runs to near completion and the water comes off close to the resistivity ceiling. It is the last polishing stage in a lot of lab and electronics systems, sitting after RO and often after EDI to take the resistivity the final distance.

Resin runs out. Every ion the resin captures uses up a site, and when the sites are full the resin is exhausted and the water quality falls off a cliff, fast, not gradually. You see it as a sharp drop on the resistivity meter at the outlet. That meter is the whole alarm system for a DI bed, which is why the inline monitor downstream of the resin is not optional. Watch the resistivity, and the resin tells you itself when it is done.

Two ways to deal with exhaustion. A regenerable system rinses the resin with acid and caustic on site to strip the captured ions and recharge the beads, which suits larger plants that can handle the chemicals. A service-exchange or service-DI setup skips the chemicals entirely: when the tank exhausts, a supplier swaps it for a fresh, regenerated tank and takes the spent one away to recharge offsite. Service exchange is common for labs and smaller users because there are no regeneration chemicals to store or handle on site, and the tradeoff is paying per tank instead of per chemical.

The RO plus DI train

The standard high-purity setup is RO followed by DI, and the architecture is the same whether the building is a clinic or a fab: pretreatment, then RO for the bulk removal, then a polishing stage to finish. The pretreatment protects the RO, the RO carries the dissolved load down by 95 percent or more, and the polish closes the gap to whatever grade the application needs.

The polish takes one of a few forms. Classic mixed-bed DI tanks, regenerable or service-exchange, are the simplest and most common. Electrodeionization, EDI, replaces the chemically regenerated resin with a continuous electrical process and shows up on systems that want to run without regeneration chemicals. Many ultrapure systems stack both: RO, then EDI for continuous polishing, then a final mixed-bed to take the resistivity the last step to 18.2. The more demanding the grade, the more polishing stages you find in series.

Sizing the train means matching every stage to the peak demand and to the stage ahead of it. The RO has to make enough permeate to feed the polish and the storage, the polish has to handle the residual load at the flow the building pulls, and the storage and loop have to carry the water to the points of use without letting it sit. Undersize the RO and the storage tank runs dry under peak draw. Oversize the storage and the water sits long enough to grow bacteria. The train is balanced, not just assembled.

Electrodeionization (EDI)

Electrodeionization is continuous deionization that runs on electricity instead of regeneration chemicals. It combines ion-exchange resin, ion-selective membranes, and a DC electric field so that the field continuously pulls the captured ions out of the resin and through the membranes to a reject stream, regenerating the resin in place. The resin never fills up and never needs an acid-and-caustic recharge, because the current is regenerating it the whole time it runs.

EDI sits downstream of RO, never on its own. It needs the low dissolved-solids feed that only RO can supply, because the electrical regeneration cannot keep up with a full raw load. Fed good RO permeate, EDI produces high-purity water continuously, commonly in the 15 to 18 megohm-cm range, and an RO-plus-EDI train is a workhorse for pharma, power, and electronics water that wants high purity without a chemical room.

The appeal is what it removes from the operation. No acid and caustic to store, handle, and neutralize. No regeneration downtime, because the process is continuous. No resin disposal cycle. The tradeoff is capital cost and the demand for clean, consistent RO feed, since EDI is less forgiving of a poorly performing RO upstream than a mixed-bed tank would be. Where the water has to be high purity and the operation does not want regeneration chemicals on site, EDI is usually the answer.

UV, ultrafiltration, and the final filter

Resistivity can read perfect while the water is still wrong, because ion-exchange and RO say nothing about organics, bacteria, or particles. The final stages handle what the purity meter cannot see, and they sit at the very end of the train, closest to the point of use.

UV does two different jobs at two different wavelengths, and mixing them up is a common mistake. UV at 254 nm is germicidal: it kills bacteria and keeps the loop biologically under control. UV at 185 nm is for total organic carbon reduction: the shorter wavelength breaks organics down into ions that the downstream polishing resin then removes, which is why a 185 nm unit usually sits ahead of a final polishing bed, not after it. Some lamps emit both. Know which wavelength you specified and what it is there to do.

After UV, the water passes a final filter, commonly a 0.2 micron membrane, to catch any bacteria or particle shed downstream of the last process, including resin fines off a polishing bed. Ultrafiltration goes further where the application demands it, removing finer particles, colloids, and bacterial endotoxin, which matters for dialysis and injectable-grade water. Ozone shows up as a sanitizing agent for storage and loops, dosed to control bacteria and then destroyed with UV before the water reaches use. These are the stages that make resistivity-clean water actually clean for the people and processes using it.

Why does pure water need a recirculating loop?

Pure water needs a recirculating loop because the instant it stops moving, it starts going bad. Still pure water is an invitation: bacteria colonize it, dissolved carbon dioxide from the air drives the resistivity down, and the very purity that makes it valuable also makes it aggressive, so it leaches contaminants out of whatever it sits against. Keep it moving and you keep it pure. Let it sit and you lose the grade you paid for.

The design that answers this is a storage tank feeding a continuous recirculating distribution loop. The pure water leaves the tank, runs out to all the points of use, and the water nobody drew off returns to the tank rather than dead-ending at a branch. A polishing stage and often a UV unit sit in the loop so the water is re-polished on every pass. The storage tank itself is protected: vented through a hydrophobic 0.2 micron vent filter or a carbon-dioxide-absorbing vent so room air does not contaminate it, or built as a bladder tank with no air contact at all.

Flow velocity in the loop is part of the design, not an accident. A common target is to keep the loop moving fast enough, often cited around 3 to 5 ft per second, that the flow is turbulent and bacteria cannot settle and build biofilm on the pipe wall. The recirculating loop is the single most important thing that separates a pure-water system that holds its grade from one that degrades between the plant and the tap. If you take one rule from distribution, it is this: the water has to keep moving.

Dead legs and stagnation

A dead leg is a branch or stub of pipe where water sits still, and it is where a pure-water distribution system goes wrong. A capped tee left from a removed outlet, an oversized run to a fixture nobody uses, a valve at the end of a long branch: any spot the recirculating flow does not reach is a pocket of stagnant water. Stagnant pure water grows bacteria and builds biofilm, and that colony seeds the rest of the loop every time the branch finally gets used.

The design rule is to keep branches short enough that the recirculating flow still scours them. A common guideline limits a dead leg to no more than a few pipe diameters off the main loop, sometimes stated as the branch length staying under about two to four times the pipe diameter, so there is no volume of water sitting out of the flow path. Points of use take off the loop with the shortest possible stub, and the loop returns to the tank rather than ending anywhere.

Dead legs are also what creeps in over a building's life. The original system was clean, then someone added an outlet, abandoned a run, or left a capped branch after a remodel, and now there is a stagnant pocket nobody designed. When a pure-water loop starts failing its bacteria counts, the abandoned branch and the capped tee are the first places to look. Find the dead leg and you usually find the contamination.

What piping is used for pure water?

Pure-water piping is plastic, and the reason is leaching. High-purity water is so depleted of ions that it pulls them out of whatever it contacts, so a piping material that gives anything up contaminates the water it is supposed to carry. Copper is the material you do not use. It corrodes and leaches into low-ion water, and dissolved copper is exactly the kind of contaminant the system spent its whole length removing. Copper belongs nowhere in a pure-water loop.

The plastics that hold up are the fluoropolymers and polyolefins. PVDF, polyvinylidene fluoride, is the high-grade choice for ultrapure and semiconductor loops: smooth bore, very low extractables, and it takes heat sanitization. Polypropylene, PP, is a common and economical choice for RO and DI distribution. PEX and other polyolefins appear on some lower-grade RO and DI runs. For the most demanding grades, high-quality electropolished stainless steel is used, particularly where the loop is hot-water or steam sanitized and a plastic would not take the temperature. The selection follows the grade: PP and PEX for general DI, PVDF for ultrapure, stainless for the high end and for heat-sanitized loops.

Joining method matters as much as the material, because a bad joint is a contamination site and a dead spot. Fluoropolymer and polypropylene systems are typically fusion-welded, not solvent-cemented with primer and glue, because solvent cements leave residue that bleeds organics into the water for a long time. The bore has to stay smooth through the fittings so there is no crevice for biofilm. Whatever you pick, the rule under all of it is the same: no copper, and nothing that leaches into water this hungry.

Sanitizing the loop

Even a well-designed loop needs periodic sanitization, because biofilm establishes slowly no matter how clean the design. Sanitization is the scheduled reset that kills the bacteria and strips the film before it builds to a level the bacteria counts catch. It is maintenance, not a fix for a bad design, and a loop that needs constant sanitizing to pass usually has a dead leg or a stagnation problem underneath.

Three methods, chosen by the system and its materials. Heat sanitization circulates hot water or steam through the loop to kill organisms, which is clean and chemical-free but needs piping and components rated for the temperature, which is part of why PVDF and stainless are chosen for those loops. Chemical sanitization circulates an agent such as peracetic acid, hydrogen peroxide, or a dilute oxidizer, then rinses it fully out, which works on any material but adds a rinse-down and a verification step. Ozone, dosed into the storage and loop and then broken down by UV before the point of use, gives continuous low-level bacterial control on systems built for it.

Whatever the method, the sanitization has a record and a schedule tied to the bacteria monitoring. You sanitize on a cadence, you verify the counts came back down, and you log it. On dialysis and pharma systems the sanitization record is part of the compliance file, not just good housekeeping, because the people downstream are patients.

Monitoring purity

A pure-water system is monitored continuously, inline, because the water quality changes faster than a periodic grab sample would catch. The core instrument is the inline resistivity or conductivity meter, temperature-compensated to 25°C, reading the water as it flows. Resistivity at the RO outlet, after the polish, and on the return loop tells you each stage is doing its job, and a resistivity drop is the first and clearest sign a polishing bed has exhausted or an RO is slipping.

Resistivity is necessary but not sufficient, because it only sees ions. The higher-grade systems add total organic carbon monitoring, inline TOC analyzers that catch the organics the resistivity meter is blind to, which is how pharma and electronics systems watch the contaminant that does not move the conductivity. Bacterial monitoring is by periodic sampling and culture, on a schedule, because there is no cheap real-time bacteria meter, and dialysis and pharma have defined sampling regimes and limits.

The monitoring earns its keep through alarms and trends, not just a number on a screen. Set alarms on the resistivity floor, the TOC ceiling, and the RO performance so a slipping stage announces itself before the water reaches a process out of spec. Trend the numbers over weeks and you see the membrane fouling or the resin approaching exhaustion in time to act on a planned change instead of a failure. The gauges are how you run the system on data instead of on a replacement calendar.

Dialysis and medical pure water

Dialysis water is held to the strictest practical standard among the common applications, because the water mixes with dialysis concentrate and contacts the patient's blood across the dialyzer membrane. Whatever is in the water goes into the patient. That single fact drives the whole design toward redundancy and tight microbial control, well past what a lab or a boiler plant would build.

The governing standards are the AAMI and ISO 23500 series for the quality of water and fluids used in hemodialysis, which set chemical contaminant limits, bacterial limits, and bacterial endotoxin limits, plus the monitoring and management to hold them. A dialysis water room is typically RO-based, often double-pass RO, with carbon ahead of it that has to be sized and tested specifically for chloramine removal because chloramine in dialysis water hemolyzes red cells. Ultrafiltration commonly sits near the point of use to hold the endotoxin limit, and the distribution is a recirculating loop on the same no-dead-leg, sanitizable principles as any pure-water loop, with the documentation a clinical setting requires.

Hospitals and surgery centers run several of these high-purity and life-safety water and gas systems side by side, and the medical gas piping that shares those same facilities is its own certified discipline, covered in the medical gas field guide. If you work pure water in a health care setting, treat the application standard and the facility's own protocols as controlling, and verify the current edition, because medical standards are revised and the limits are not yours to estimate.

Lab, pharma, and semiconductor water

These three applications share the same train and split on the grade and the standard they answer to. All of them run RO into a polish into a recirculating loop. What differs is how pure the polish has to get and which limits the system is built to pass.

Lab water is graded by the ASTM D1193 Type I, II, III scheme, and a lab usually wants Type I ultrapure near 18 megohm-cm at the bench for sensitive analysis like HPLC, ICP-MS, and PCR, with Type III RO-grade for glasswashing and feed. The point-of-use polisher at the bench is what guarantees the Type I, because the resistivity falls between the plant and the tap. Pharmaceutical water answers to the USP compendial grades, Purified Water and Water for Injection, which are defined by conductivity and total organic carbon limits, with WFI carrying a bacterial endotoxin limit on top, so a pharma system is built and validated around those compendial limits rather than a single resistivity number.

Semiconductor and electronics water is the most demanding, the ultrapure water that rinses wafers, where a single dissolved-solid particle bridging a circuit feature kills the chip. These systems run the full train, RO, EDI, mixed-bed polish, UV, and ultrafiltration, and they chase the 18.2 megohm-cm ceiling along with extremely low TOC, particle, and bacteria counts. Across all three, the rule is the same: build to the application's standard, verify the current edition, and let that standard set the grade rather than a habit carried from another job.

Boiler feed, humidification, and data centers

Not every pure-water job is about a lab result. A large share of it is about keeping scale off equipment, and for that the grade is lower but the payoff is direct: the dissolved solids that would bake onto a surface never get there in the first place.

Boiler feed is the classic case. Hardness and silica in makeup water precipitate on hot boiler tubes as scale, the scale insulates the tube from the water that is supposed to cool it, and the overheated metal fails. Feeding the boiler RO or DI water removes the minerals that form the scale, which protects the tubes, holds the heat-transfer efficiency, and cuts the blowdown needed to control dissolved solids. Steam and evaporative humidifiers have the same problem in miniature: hard water crusts the electrodes, cylinders, and nozzles, and pure-water feed keeps them clean and the mineral dust out of the conditioned air.

Data centers run pure water for the same scale reason, in evaporative and adiabatic cooling and in humidification, where mineral-laden water would scale the media and dust the white minerals into the airstream and onto equipment. The grade for these duties is usually RO or RO-plus-DI rather than full ultrapure, because the goal is no scale and no dust, not a Type I resistivity. Match the grade to the duty: scale control does not need 18.2 megohm-cm, and paying for ultrapure where RO would do is money spent on a number nobody downstream uses.

Maintenance and upkeep

A pure-water system is not a set-and-forget install. It is a chain of consumables that wear out on different schedules, and the water quality holds only as long as every link is maintained. The good news is that the maintenance is predictable, and the monitoring tells you when most of it is due.

The consumables, roughly by how often they come due: sediment and carbon prefilters on a regular change interval so the carbon keeps stripping chlorine and the sediment filter keeps protecting the membrane; the RO membrane cleaned on a CIP schedule and replaced when cleaning no longer recovers it; the DI resin regenerated or the service tank swapped when the resistivity drops; the UV lamp replaced on its rated hours, because a lamp still glowing past its rated life has lost the output that does the work; and the final and vent filters changed on schedule. Sanitization of the loop runs on its own cadence alongside all of it.

Tie the schedule to the monitoring and to records, not to memory. Resistivity trending shows the resin and membrane approaching the end before they fail. Pressure-drop trending shows filters loading up. A maintenance log that captures what was changed, when, and what the gauges read at the time is what lets the next person run the system instead of rediscovering it. The systems that fail early are almost never bad equipment. They are good equipment whose carbon, membranes, and resin nobody kept up.

What to document

A pure-water system lives or dies on its records, because the quality is invisible and the failures are gradual. The log is what tells you whether today's resistivity drop is normal drift or a resin about to quit, and on dialysis and pharma systems the record is a compliance requirement, not a courtesy. Capture each stage, what it does, and the number that says it is working.

Common mistakes

• No pretreatment, so chlorine oxidizes the RO membrane and hardness scales it, and membranes fail in months.
• Dead legs and stagnant branches that grow bacteria and seed the loop, often abandoned runs left after a remodel.
• Copper or another leaching material in the loop, contaminating water that is hungry enough to pull ions out of the pipe.
• No recirculating loop, so the stored or distributed water sits, loses resistivity to carbon dioxide, and grows bacteria.
• Ignoring a resistivity drop, which is the resin telling you it is exhausted, until the water has been out of spec for a while.
• No sanitization schedule, so biofilm builds in the loop until the bacteria counts fail.
• Building the wrong grade for the application, paying for ultrapure where RO would do, or running DI where the standard demands validated USP or AAMI quality.

Field checklist

Standards and references

Pure water is governed by the application's standard, not by a single plumbing code, so name the right one for the job and verify its current edition. For laboratory reagent water, ASTM D1193 defines the Type I, II, III, IV grades by resistivity, total organic carbon, and other limits. For pharmaceutical water, the USP compendial monographs for Purified Water and Water for Injection set conductivity and total organic carbon limits, with the conductivity and TOC test methods in their own USP chapters. For dialysis, the AAMI and ISO 23500 series set the chemical, bacterial, and endotoxin limits and the quality management around them.

Above the standard sits the equipment and the project specification. The analyzer, the dialysis machine, the boiler, or the fab tool can demand a purity, a flow, and a pressure that the general grade does not capture, and the equipment manufacturer's requirement controls where it is stricter. The recirculating-loop and no-dead-leg design principles run through high-purity practice across all of these because the physics of stagnation is the same regardless of which standard the water answers to.

Three things do not bend across any of it. Pretreat to protect the RO membrane, because chlorine and hardness destroy it. Keep the water moving in a recirculating loop with no dead legs, because still pure water degrades and grows bacteria. And keep copper out, because water this depleted leaches metals out of the pipe. Beyond those, hedge the grade and the resistivity to the standard and the application, confirm the adopted edition, and let the project specification override any rule of thumb when it is stricter.

Units, terms, and conversions

Pure-water work crosses a few unit systems, and the same purity reads differently on a lab spec, a pharma monograph, and a meter, so know the equivalents.

Purity shows up as resistivity in megohm-cm, where higher is purer and ultrapure tops out near 18.2 at 25°C, or as conductivity in microsiemens per cm, the inverse, where lower is purer and that same water reads about 0.055. TDS in parts per million is the plain read on dissolved load, often estimated from conductivity. Organics are measured as total organic carbon, TOC, usually in parts per billion. All electrical readings are referenced to 25°C, so a meter without temperature compensation is not giving you a comparable number.

RO / permeate / reject

Reverse osmosis; the purified water through the membrane is permeate, the concentrated leftover is reject or concentrate

DI / deionization

Ion-exchange polishing that trades dissolved ions for hydrogen and hydroxide on resin

EDI

Electrodeionization, continuous DI regenerated by an electric field, no regeneration chemicals

Mixed bed

Cation and anion resin blended in one vessel for the highest-purity polish

Recovery

The fraction of RO feed water that becomes permeate rather than going to reject

Dead leg

A branch or stub where water sits out of the recirculating flow and grows bacteria

TOC

Total organic carbon, the organics measure that resistivity is blind to, usually in parts per billion

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• How does a water softener work?
• Who can install medical gas piping?
• What is the difference between upfeed and downfeed water distribution?
• What is a corporation stop?
• Why is my water pressure low?