4.3 Membrane Filtration: Microfiltration, Ultrafiltration, Nanofiltration & Reverse Osmosis

Membrane Separation Spectrum and Classifications

Membrane filtration utilizes semipermeable synthetic barriers to physically separate dissolved and suspended constituents from water under an applied pressure differential. In modern drinking water treatment, membrane technologies can provide a compact, directly testable particulate-pathogen barrier. Performance depends on membrane selection, pretreatment, integrity and operation; no membrane should be described as an infallible barrier.

Membrane processes are classified into four distinct operational regimes based on nominal pore diameter, Molecular Weight Cut-Off (MWCO, measured in Daltons, $\text{Da}$), and operating pressure:

Flow Regimes and Module Configurations

1. Hydraulic Flow Configurations

• Dead-End Filtration: $100%$ of feed water is driven perpendicular through the membrane barrier, converting the entire volume into filtrate (permeate). Retained solids accumulate directly on the membrane surface as a cake layer. Dead-end mode is energy-efficient and standard for low-pressure hollow-fiber MF/UF, relying on periodic automated air-water backwashes every $15\text{ to }60\text{ minutes}$.
• Cross-Flow (Tangential) Filtration: Feed water is pumped tangentially across the membrane surface at high velocity. The feed stream splits into two streams: purified permeate ($75%\text{ to }85%$) that passes through the membrane, and a concentrated reject (concentrate/brine) stream ($15%\text{ to }25%$) that sweeps accumulated solutes away from the membrane face. Cross-flow is the usual NF/RO configuration used to control concentration polarization and carry rejected solutes to the concentrate stream.

2. Physical Module Designs

• Hollow-Fiber Modules: Bundles of thousands of microscopic hollow polymer capillaries (PVDF, PES, or polysulfone) potted in resin. Water flows either outside-in (feed on outer shell, filtrate drawn from inner lumen; easy to clean, handles high solids) or inside-out (feed pumped through lumen; uniform hydraulics). Hollow fibers are fully backwashable and dominate municipal MF/UF surface water plants.
• Spiral-Wound Elements: Flat-sheet membrane envelopes separated by porous permeate carriers and mesh feed spacers wound around a central perforated permeate collection tube. Spiral-wound elements provide exceptionally high surface area packing density but cannot be hydraulically backwashed (solids lodge in mesh spacers). Standard for NF and RO.

3. Vacuum-Immersed vs. Pressure-Driven Skid Vessels

• Vacuum-Driven (Immersed / Submerged): Hollow-fiber membrane cassettes are submerged directly in open concrete process tanks. Permeate is drawn inward through fiber walls under negative pressure (vacuum of $-3\text{ to }-12\text{ psig}$) generated by suction pumps. Allows direct visual inspection, coarse aeration for scouring, and lower operating energy.
• Pressure-Driven Skids: Membrane modules are enclosed in cylindrical fiberglass reinforced plastic (FRP) pressure vessels mounted on skids. High-pressure pumps drive feed water into vessels at $30\text{ to }1,000\text{ psig}$.

Key Membrane Operating Parameters and Hydraulic Calculations

Plant operators monitor membrane health, hydraulic productivity, and fouling rates using standardized parameters:

1. Membrane Flux ($J$)

Flux is the volumetric filtration rate of permeate produced per unit of active membrane surface area, expressed in gallons per square foot per day (GFD) or liters per square meter per hour (LMH): $J = \frac{Q_p}{A_m}$ Where:

• $Q_p$ = Permeate flow rate (Gallons per Day, GPD)
• $A_m$ = Total active membrane surface area ($\text{ft}^2$)
• Typical Design Values: $20\text{ to }45\text{ GFD}$ for low-pressure MF/UF; $10\text{ to }20\text{ GFD}$ for NF/RO.

2. Transmembrane Pressure (TMP)

Transmembrane Pressure (TMP) represents the average net pressure differential driving water molecules across the membrane barrier: $\text{TMP} = \left( \frac{P_{inlet} + P_{outlet}}{2} \right) - P_{filtrate}$ Where:

• $P_{inlet}$ = Feed pressure at module inlet (psig)
• $P_{outlet}$ = Concentrate pressure at module outlet (psig)
• $P_{filtrate}$ = Permeate pressure on clean side (psig)

For vacuum immersed membranes: $\text{TMP} = P_{basin_static} - P_{permeate_vacuum}$.

3. Temperature Correction and Specific Flux

Because water viscosity increases by approximately $2.5%\text{ for every }1^\circ\text{C}$ drop in water temperature, measured flux must be normalized to standard temperature ($20^\circ\text{C}$ or $25^\circ\text{C}$) to distinguish temperature-induced resistance from true chemical fouling: $J_{20^\circ\text{C}} = J_T \times \left( \frac{\mu_T}{\mu_{20^\circ\text{C}}} \right)$ $\text{Specific Flux} = \frac{\text{Normalized Flux (GFD)}}{\text{TMP (psi)}}$ Trend temperature-normalized permeability or specific flux against the manufacturer’s and facility’s approved cleaning trigger; a fixed percentage decline is not universal.

4. Recovery Rate ($R$) and Concentration Factor ($CF$)

$\%\text{ Recovery} = \left( \frac{Q_{permeate}}{Q_{feed}} \right) \times 100\%$ $\text{Concentration Factor (CF)} = \frac{1}{1 - (\text{Recovery}/100)}$ Example: A nanofiltration system operating at $80%$ recovery ($R = 0.80$) concentrates rejected minerals by a factor of $CF = 1 / (1 - 0.80) = 5.0\times$ in the brine stream.

Membrane Fouling Pathology and Clean-in-Place (CIP) Regimes

Membrane fouling increases TMP, drives up energy consumption, and degrades permeate quality. Fouling is classified into four operational categories:

1. Particulate / Colloidal Fouling: Clays, silts, and metal hydroxides forming a cake layer on membrane surfaces.
2. Organic Fouling: Adsorption of natural organic matter (NOM), humic and fulvic acids, biopolymers, and proteins onto polymer pores.
3. Mineral Scaling: Precipitation of sparingly soluble salts ($CaCO_3, CaSO_4, BaSO_4, SrSO_4$, and silica $SiO_2$) exceeding their solubility limits in the concentrate stream of NF/RO units.
4. Biofouling: Attachment and growth of living bacterial biofilms and secretion of extracellular polymeric substances (EPS).

Clean-in-Place (CIP) Chemical Protocols

When standard backwashes or maintenance washes fail to restore permeability, a full Clean-in-Place (CIP) may be required. The table gives representative chemistries—not universal recipes. Operators must follow the membrane manufacturer’s approved concentration, pH, temperature, exposure-time, neutralization and waste-disposal limits:

Membrane Integrity Testing: Direct vs. Indirect Protocols

Under the EPA Long Term 2 Enhanced Surface Water Treatment Rule (LT2), a membrane used for Cryptosporidium treatment credit receives the lower of (1) removal demonstrated by product-specific challenge testing or (2) the log-removal value the direct integrity test can verify during operation. The rule does not assign one default credit from membrane type or nominal pore size.

1. Direct Integrity Testing (High Sensitivity)

A qualifying direct integrity test must meet the approved resolution, sensitivity and frequency criteria. LT2 generally requires a resolution of 3 µm or smaller and testing at least once each day the membrane unit is in operation, although the state may approve less frequent testing under the rule’s reliability provisions:

• Pressure Decay Test (PDT): The isolated, wetted module is pressurized with clean air below the intact membrane’s bubble point, the source is isolated, and pressure loss is monitored for the approved test duration.
  • Pass/Fail criterion: An integral unit has little diffusive loss. Failure occurs when the measured decay exceeds the state- and manufacturer-approved control limit for that membrane unit and test setup; 0.1–0.3 psi/min may appear in example protocols but is not an EPA-wide limit.
  • Resolution: Calculates the Log Removal Value (LRV) verified by the test: $\text{LRV} = \log_{10}\left(\frac{Q_p \times C_{feed}}{Q_{breach} \times C_{filtrate}}\right)$
• Diffusive Air Flow Test: Measures the volumetric flow rate of air diffusing through wetted membranes at elevated test pressure.
• Bubble Point Test: Determines the critical pressure required to force air through the largest wet pore ($P = \frac{4\gamma \cos\theta}{d}$). Used to test flat-sheet integrity.

2. Indirect Integrity Monitoring (Continuous)

• Online Permeate Turbidimetry: Continuous turbidimeters on each skid permeate header recording data every 15 minutes. Compare the signal with the approved indirect-integrity control limit and investigate a significant excursion; no single 0.05–0.10 NTU band applies to every membrane installation.
• Laser Particle Counters: Measures particles in the $2\text{ to }5\text{ }\mu\text{m}$ range. A sudden spike in particle counts indicates a fiber break before turbidity rises.

3. Fiber Pinning and Repair

When a Pressure Decay Test fails, operators locate the compromised fiber using sonic acoustic detectors or sight-glass bubble inspection during low-pressure air injection. The damaged fiber is isolated or the module repaired using the manufacturer-approved method, which may include pinning or plugging both lumen ends. The unit must pass its direct integrity test before return to credited service.