A membrane system almost never fails suddenly — it drifts. Scale, fouling, and exhausted pretreatment build so gradually that no single visit shows a problem. That's the whole case for PM: same readings, same way, every visit, so the drift shows up on paper months before it shows up as a dead system and an emergency call.
One visit's reading — meaningless alone The trend — the real story Action threshold
Every individual reading looked "fine." The trend caught the fouling four months before it became a failure.
Video overview
8:40
Narrated overview of this module. The article below has the full procedure and the numbers.
01
Why PM matters this much
Membrane damage is one-way past a point
Fouling cleans off easily when it's caught early — clean at a 10–15% drop in normalized permeate flow (or a similar rise in ΔP or salt passage). Let it run further and foulant compacts, scale anchors, and some of that loss becomes permanent. PM is what catches the drop at 10% instead of 40%.
Planned beats emergency, every time
A trend that says "clean next month" means a scheduled visit and zero customer downtime. The same problem discovered by a "system's down" call means after-hours labor, rush parts, an unhappy customer — and often a membrane replacement instead of a cleaning.
Warranties and disputes run on records
Membrane warranties expect documented operating data and functioning pretreatment. Our PM logs and photos are the proof that the system was run right — protecting Axis and the customer if hardware fails or a claim ever gets made.
Pretreatment failures are silent
A softener that quit regenerating or a carbon bed passing chlorine doesn't alarm — it just quietly destroys the membranes downstream. The only way anyone finds out in time is a tech testing each barrier on every visit.
02
What to test and record — every visit, no exceptions
The value is in the consistency: same points, same instruments, same order. A reading you skip is a hole in the trend.
ReadingHowWhat drift means
Pressures
Feed pressureat membrane inlet
panel gauge / HMI
creeping up → fouling or scaling building resistance
Interstage & concentrate pressuremulti-stage systems
panel gauges
ΔP rising per stage points to where the problem is — front = fouling, tail = scale
Cartridge filter ΔPinlet vs outlet
both gauges
rising → filters loading; change on ΔP, not the calendar alone
Flows
Permeate flow
flow meter
falling (normalized) → fouling/scale; the headline number
Concentrate flow+ recycle if fitted
flow meter
confirms recovery hasn't drifted from its set point
Recoverypermeate ÷ (permeate + concentrate)
calculated
creeping up unnoticed = scaling risk — see the recovery article
Quality
Feed & permeate conductivity/TDS
meter / panel
calculate % rejection; rising salt passage → scale, damage, or o-ring leak
Feed pH
calibrated pH meter
shifts change scaling potential & chemical demand
Temperature
panel / thermometer
required to normalize flow — see the next section
Pretreatment & chemicals
Hardness after softener
titration kit
any hardness through = regen problem — membranes at risk
Free chlorine after carbon / SBS
DPD test
must be zero ahead of the membranes — non-negotiable
Chemical day-tank levelsantiscalant, hypo, SBS
sight / measure
actual use vs. expected proves the pumps are really dosing
Pump drawdown checkwhen due, or if use looks off
graduated cylinder
verifies ml/min — see the antiscalant & chlorine articles
03
Recording the data so it can actually tell the story
Numbers on a clipboard in the truck are worthless. Every PM's readings go into the service report on the Salesforce work order, same day — that's what makes site history searchable, comparable, and safe from a lost notebook. Three habits make the data trustworthy:
Log against a baseline
Every system's log starts from its commissioning baseline — the flows, pressures, rejection, and recovery recorded when it started up clean. Every later reading is judged against that baseline, not against memory or last month alone.
Normalize before you judge flow
Permeate flow moves roughly 3% per °C of feed temperature all by itself — cold winter feed water can make a perfectly clean system "look" fouled, and warm summer water can hide real fouling. That's why temperature gets recorded every visit: flow is compared on a temperature-normalized basis (the data sheet / software does the correction) before anyone declares a trend.
The action triggers, referenced to baseline at the same temperature: normalized permeate flow down 10–15% · ΔP up ~15% · salt passage up noticeably. Any one of these means schedule a cleaning or investigation — on our schedule, not the failure's.
Photograph everything that matters
Photos are the part of the record a number can't capture — attach them to the same work order:
Every visit
The panel
Gauges, flow meters, and HMI screen exactly as found — a timestamped snapshot of system state.
Every change
Filters, out of the housing
The used cartridges before they hit the trash. Color and loading pattern diagnose the upstream water.
Every visit
Day tanks & salt
Chemical levels and brine-tank salt — proves consumption and backs up the level readings.
Always
Anything abnormal
Leaks, corrosion, staining, damaged insulation, unsafe conditions — photograph first, then fix or flag.
Before / after
Any repair you make
The condition you found and the condition you left. Ends "who touched it last" disputes before they start.
Once per site
Nameplates & layout
Equipment nameplates and a wide shot of the room — the next tech (or the office) troubleshoots remotely with these.
04
Changing cartridge filters — the right way
Cartridge filters are the membranes' last bodyguard. Change on differential pressure — typically when ΔP rises ~8–10 psi over the clean reading — or on the site's schedule, whichever comes first. A filter that "still looks okay" at high ΔP is channeling and passing solids.
Record the loaded ΔP first. That number is data — how fast filters load at this site sets the right change interval.
Isolate and depressurize the housing. Close isolation valves, bleed pressure at the vent — never crack a pressurized housing.
Pull the cartridges and photograph them. Even brown loading is normal duty. Slime = biofouling upstream. One side loaded = flow problem. Orange/brown slick = iron getting through.The used filter is a free water-quality report — read it before you toss it.
Inspect and wipe the housing; check the o-ring. Lube the o-ring with clean silicone grease (never petroleum on membranes' feed side), replace if flattened or nicked.
Install new cartridges with clean hands/gloves. Handle by the ends — a contaminated filter seeds the housing with the bacteria it's supposed to stop. Confirm correct micron rating.
Refill, vent the air, return to service, and record the clean ΔP plus the change date and cartridge type on the work order.Clean ΔP is the new reference for "when's the next change."
05
Proving the pretreatment is working — barrier by barrier
Each pretreatment stage protects the membranes from one specific threat, and each one has a simple pass/fail test. "The equipment is running" is not the test — the water leaving it is.
Sediment / cartridge filtrationstops particles from plugging the lead elements
Test: ΔP across housingFail: ΔP high, or solids visible downstream
Water softenerstrips hardness so CaCO₃ can't scale the membranes
Test: hardness titration at outlet ≈ 0 gpg + salt in brine tank + regen clock rightFail: any hardness through
Carbon filter / SBS injectionremoves chlorine before it can oxidize the membranes
Test: DPD free chlorine at outlet = 0.0Fail: any detectable chlorine — stop and fix before it reaches the RO
Antiscalant feedholds supersaturated salts in solution through the array
Test: day-tank drop matches expected use + drawdown when dueFail: level not moving = pump not pumping
Instrumentation itselfbad gauges make good systems look sick — and sick ones look fine
Test: gauges read ~0 when depressurized; meters cross-check against bucket tests; pH/TDS meters calibratedFail: trust nothing until fixed
Order matters when diagnosing: if RO numbers drift, check the pretreatment tests first. A softener passing hardness or carbon passing chlorine explains the drift — and cleaning the membranes without fixing the barrier just buys a few weeks before it happens again.
06
The visit, start to finish
The working checklist — tap items off as you go. (Resets on reload; the permanent record is the Salesforce work order.)
Standard PM visit checklist
In order — the sequence is designed so pretreatment is verified before RO numbers are judged.
✓Walk the system as-found. Photograph the panel and anything abnormal before touching a thing.photos → work order
✓Verify pretreatment barriers. Hardness after softener · free chlorine after carbon = 0.0 · cartridge ΔP · salt level.titration + DPD + gauges
✓Check chemical feeds. Day-tank levels vs. expected use; drawdown test if due or if consumption looks off.graduated cylinder
✓Record the full panel. Pressures (feed / interstage / concentrate), flows (permeate / concentrate), recovery, feed & permeate TDS, pH, temperature.every point, every visit
✓Change filters if ΔP or schedule says so. Photograph the used cartridges; record loaded and clean ΔP.see filter procedure above
✓Compare against the trend, not just today. Normalized flow, ΔP, and rejection vs. baseline — flag anything past the action triggers.−10–15% flow = schedule clean
✓Housekeeping & leaks. Tighten, wipe, label; note anything the customer should know about.
✓Log it all in Salesforce before leaving the site. Readings, photos, filters changed, chemicals added, and recommendations on the work order.same day, no exceptions
0 of 8 done
A PM where nothing was wrong isn't a wasted trip — it's a data point that proves the trend is flat. That is the product.
The one thing to remember
One reading proves nothing. The trend catches the failure months early — but only if every visit feeds it the same data, recorded the same way.
Axis Water Technologies · 855-443-2947 · info@axiswater.com · axiswater.com · 3462 W Loop 289, Lubbock, TX 79407Axis technician training library · content as of October 1, 2026
