2.8 Membrane Systems: MF, UF, NF and RO Operation & Integrity Testing

Membrane Systems: MF, UF, NF and RO Operation & Integrity Testing

Virginia's waterworks classification protocol names "chemical coagulation in combination with membrane filtration" in Class 1, Class 2, and Class 3, and "membrane or other filtration technologies without chemical coagulation" in Class 3 and Class 4. Membrane plants are now mainstream, and the ABC outline includes membrane filtration under both Treatment Process and Equipment Operation and Maintenance.

---

1. The Separation Spectrum

MF and UF are absolute physical barriers to Giardia and Cryptosporidium, which is why membrane plants earn high log-removal credit - but that credit is contingent on demonstrated integrity, not on the nameplate pore size.

Configurations

• Hollow fiber - the dominant low-pressure format. Flow may be outside-in (feed on the shell side, permeate drawn inside the fiber - more tolerant of solids) or inside-out (feed pumped down the lumen - easier to clean by flushing but plugs on debris).
• Pressurized vessels versus submerged/vacuum racks. Submerged systems draw permeate under vacuum, tolerate higher solids, and use less energy; pressurized systems achieve higher flux and are easier to integrity-test.
• Spiral wound - the standard for NF and RO, with feed spacers and permeate carriers wound around a central tube. Spiral elements are intolerant of particulates, so they always require MF/UF or cartridge pretreatment.

---

2. The Core Operating Metrics

Flux

Flux is permeate flow normalized to membrane area, in gallons per square foot per day (gfd):

Flux = Permeate flow / Membrane area

Typical low-pressure design flux is 25 to 75 gfd on good-quality feed, lower on high-solids water.

Transmembrane pressure (TMP)

In dead-end operation:

TMP = Feed pressure - Permeate pressure

In crossflow operation, feed and concentrate pressures differ, so the average feed-side pressure is used:

TMP = [(Feed pressure + Concentrate pressure) / 2] - Permeate pressure

Specific flux (permeability)

This is the metric that actually tells you the membrane's condition, because it removes both flow and pressure from the comparison:

Specific flux = Flux / TMP, in gfd per psi

Because water viscosity falls as temperature rises, specific flux must be temperature-corrected to a reference temperature (usually 20 degrees C) before trending. An uncorrected specific flux drops every autumn and recovers every spring purely from viscosity, which masks real fouling.

Recovery

Recovery (%) = (Permeate flow / Feed flow) x 100

Low-pressure membranes commonly run 90 to 97 percent recovery. RO commonly runs 70 to 85 percent in a two-stage array. Raising recovery concentrates the salts in the reject stream, which is exactly what drives scaling.

---

3. Fouling: Diagnosis by Signature

Fouling is any accumulation that reduces specific flux. The diagnostic question is always: does it recover after a backpulse, and does it recover after a chemical clean?

Polyamide RO and NF membranes are destroyed by free chlorine. Feed water to a polyamide element must be dechlorinated - typically with sodium bisulfite or granular activated carbon - and the ORP of the feed monitored continuously as an alarm. Cellulose acetate membranes tolerate low chlorine but hydrolyze outside pH 4 to 6.

---

4. Cleaning: Backpulse versus CIP

A chemically enhanced backwash is a routine backpulse dosed with a low concentration of hypochlorite (for organics and biofilm) or acid (for metal oxides), usually daily. It extends the interval between full CIPs.

The CIP decision rule: clean when temperature-corrected specific flux drops to a defined percentage of the post-CIP baseline - commonly 15 to 20 percent - rather than on a fixed calendar. Cleaning too often shortens membrane life; cleaning too late lets fouling become irreversible.

---

5. Integrity Testing

Log removal credit for Cryptosporidium depends on proving the barrier is intact. Two monitoring layers are used.

Direct integrity testing (DIT)

• Performed at least once per day on each unit.
• The standard method is a pressure decay test (PDT): the module is drained on one side, pressurized with air (typically 15 to 40 psi, above the bubble point of the largest defect that must be detected), isolated, and the pressure decay measured over a fixed interval (usually 5 to 10 minutes).
• The test must have a resolution sufficient to detect an integrity breach of 3 micrometers or less, must have adequate sensitivity to verify the log removal value being claimed, and must have an established control limit. Exceeding the control limit requires taking the unit offline, finding and pinning the broken fiber, and retesting.
• Related methods include the diffusive airflow test and the marker-based (particulate or molecular) challenge test.

Continuous indirect integrity monitoring (CIIM)

• Filtrate turbidity monitored at least every 15 minutes on each unit, with a typical action level of 0.15 NTU for more than 15 minutes.
• Particle counting or laser turbidimetry provides earlier warning than nephelometric turbidity because membrane filtrate turbidity is normally so low that ordinary turbidimeters sit near their detection floor.

Indirect monitoring does not replace the daily direct test; it fills the interval between direct tests.

---

6. RO-Specific Operations

Normalization

Raw RO readings are meaningless for trending because feed temperature, feed pressure, feed TDS, and recovery all move. Operators track normalized permeate flow, normalized salt passage, and normalized differential pressure, all corrected to a reference condition. The industry rules of thumb for when to clean:

• Normalized permeate flow down 10 to 15 percent;
• Normalized salt passage up 5 to 10 percent; or
• Stage differential pressure up 15 percent.

Scaling control

The concentrate stream is where saturation is reached. Control levers:

1. Antiscalant - threshold inhibitors dosed at 2 to 5 mg/L that keep CaCO3, CaSO4, BaSO4, and silica supersaturated without nucleating.
2. Acid addition - sulfuric or hydrochloric acid to convert bicarbonate to carbonic acid, lowering the Langelier Saturation Index of the concentrate. Note that sulfuric acid adds sulfate and can worsen CaSO4 or BaSO4 risk.
3. Lower recovery - the direct lever. Dropping recovery from 80 to 75 percent meaningfully reduces concentrate saturation.
4. Softening pretreatment - lime softening or weak-acid cation exchange upstream.

Post-treatment

RO permeate is aggressive: near-zero alkalinity and hardness, low pH, and a strongly negative LSI. It must be stabilized before distribution - by blending with a bypass stream, by calcite (limestone) contactors, or by adding lime or caustic plus carbon dioxide - or it will attack cement mortar linings and metallic plumbing.