Tertiary Ultrafiltration of Secondary Effluent for Reuse: Screening, Inline Coagulation, Flux, Fouling, Cleaning and Cost

What the UF has to deal with

Secondary effluent from a well-run activated sludge plant typically carries 10 to 20 mg/L suspended solids, 40 to 60 mg/L COD of which most is soluble, 5 to 10 mg/L BOD, 5 to 10 mg/L orthophosphate where there is no chemical dosing, and a colloidal fraction of microbial polymers (soluble microbial products and extracellular polysaccharides, together with humic matter) at 7 to 10 mg/L as dissolved organic carbon. Solids and hair clog fibres and are dealt with by screening. Colloidal biopolymers are what fouls the membrane: they are too large to pass a 0.02 µm pore, too small to backwash cleanly, and they form a compressible gel layer that is only partly reversed by hydraulic cleaning. Their quantity varies with the upstream plant: short secondary hydraulic retention and low operating temperature both raise the biopolymer load, and recent pilot work has shown that the secondary HRT and temperature interact in setting tertiary fouling rates and the preferred coagulant dose. Seasonal change is therefore a property of the specific works, not a general rule; some sites see their worst fouling in warm months from algal and bacterial regrowth in effluent channels, others in winter from viscosity and sludge washout after storms.

Pre-screening

Hollow-fibre manufacturers require feed screened to 100 to 300 µm, with inside-out multibore fibres at the finer end because a hair lodged in a 0.9 mm capillary is not removed by backwash. A self-cleaning disc or basket strainer at 100 to 150 µm is the minimum. A microscreen drum filter at 20 to 40 µm ahead of the strainer is worth considering where effluent solids exceed about 15 mg/L or where the secondary clarifiers are prone to washout: pilot work combining a drum filter at 1.7 to 6.6 m³/m²·h with UF showed the drum removing most of the suspended solids and turbidity and extending UF cycle times, and a drum filter also protects the membrane against a clarifier upset that would otherwise force a shutdown. Either way the screen must be upstream of the coagulant injection point, or the flocs will be removed before they reach the membrane and the coagulant wasted.

Inline coagulation

Dosing a few mg/L of iron or aluminium directly into the feed line, with no flocculation tank and no settlement, is the single most effective fouling control available. Pilot trials on activated sludge effluent at a municipal works in Israel compared 0, 1, 5, 10 and 15 mg/L Fe³⁺ through 50 kDa, 100 kDa and 0.2 µm multibore modules at 60 L/m²·h with 30 to 45 minute cycles and a 56 second backwash at 230 L/m²·h. One milligram per litre of iron prevented irreversible fouling as effectively as 5 or 10 mg/L; the higher doses produced more sludge and higher TMP within the cycle without lowering the post-backwash baseline. Conventional doses of 10 to 30 mg/L FeCl₃ (3.5 to 10 mg/L Fe) quoted for separation-oriented pretreatment are therefore several times more than needed when the aim is only to protect the membrane. Dosing only during the first 2.5 minutes of each filtration cycle, when the dynamic cake is forming, matched continuous dosing in the same trials and cut coagulant consumption by 94 percent: once an initial floc layer is in place it traps fresh foulants itself.

Two consequences follow for design. Iron hydroxide on the membrane is an inorganic foulant and must be removed by an acid step in the cleaning regime, so an acid CEB is not optional on a coagulated plant. And the orthophosphate in the feed is partly precipitated, which is useful for a reuse quality target but means the dose must be checked against the phosphate present, since at 7 mg/L PO₄-P the 1 mg/L Fe is stoichiometrically consumed many times over and the fouling benefit comes from the floc rather than from phosphate removal. The iron chemistry is the same as in chemical phosphorus removal, but the objective and dose are different.

Flux, temperature and viscosity

Design fluxes for secondary effluent on pressurised dead-end hollow fibres fall in the range 50 to 80 L/m²·h gross, with 60 L/m²·h the common long-term value and recovery of 86 to 95 percent depending on cycle length and backwash volume; manufacturer envelopes run to 180 L/m²·h but these are for clean surface water. Submerged air-scoured systems run at 10 to 22 L/m²·h, with a semi-industrial plant on Spanish municipal effluent sustaining 21.5 L/m²·h at about 120 mbar with a fouling rate of only 0.024 mbar per day once air intensity, solids concentration and acid backwashing were optimised. Pressurised units need less area per cubic metre and less energy, submerged units tolerate higher solids and are often the choice where the UF is a retrofit downstream of marginal clarifiers.

Flux is reported at 20 °C and the correction is viscosity:

UK secondary effluent sits between 7 and 20 °C through the year, so the winter case should fix the membrane area if output must be maintained, with the TMP limit of the module (typically 1.5 bar filtration for multibore PES) checked at the fouled permeability. Where reuse demand is seasonal, as for irrigation, the summer case may govern instead, and a seasonal operating window is reasonable: one Spanish demonstration plant set different flux and air-scour ranges for summer and winter and found energy use of 0.31 kWh/m³ in summer against 0.14 in winter because of the extra sparging needed to hold summer fouling.

Backwash, CEB and CIP

Three tiers of cleaning run on different clocks. Hydraulic backwash every 20 to 60 minutes at 1.5 to 3 times the filtration flux (230 to 300 L/m²·h on multibore fibres) for 45 to 90 seconds removes the reversible cake; adding an air scour to the backwash on pressurised systems was found to be 1.25 to 2 times as effective as water alone, allowing 60 to 80 L/m²·h with 40 to 60 minute cycles to be sustained year-round on French municipal effluent. Chemically enhanced backwash, typically once a day with hypochlorite at 100 to 200 mg/L free chlorine and caustic to pH 12, soaking 15 to 30 minutes, plus an acid step at pH 2 with sulphuric, hydrochloric or citric acid once a day to once a week on an iron-dosed plant, restores the baseline permeability. Trials of CEB reagents at 50 to 400 mg/L on effluent found acid at low pH the more effective where iron and calcium deposits dominate, with permeability recovery rising steeply with concentration. Clean-in-place, every one to three months or when permeability falls to about half of its clean value, uses the same chemistry at higher strength and longer contact, within the membrane limits (a common PES multibore product is rated for 200 mg/L free chlorine and a lifetime exposure of 200,000 mg·h/L at pH up to 13). Hypochlorite use is also a chlorinated by-product issue for the permeate and the backwash return, which goes to the head of the works and should be accounted for in the works load.

Worked example: 5,000 m³/day reuse plant

A works is to supply 5,000 m³/day of Class A quality water to an industrial user from nitrified activated sludge effluent with 15 mg/L suspended solids and a winter minimum of 9 °C.

• Screening. Duty-standby 150 µm self-cleaning strainers after a 30 µm drum filter sized at 4 m³/m²·h, with the drum backwash returned to the inlet works.
• Coagulation. 1 mg/L Fe as ferric chloride, injected after the strainers through a static mixer: 5 kg Fe/day, about 26 L/day of 40 percent solution. Intermittent dosing at the start of each cycle can be trialled after commissioning.
• Cycle and recovery. 45 minute filtration at 65 L/m²·h gross, 60 second backwash at 230 L/m²·h plus a short forward flush: permeate 48.8 L/m² per cycle, cleaning water about 5.8 L/m², recovery 88 percent before CEB and CIP losses.
• Area. Feed to the membranes 5,000 / 0.88 = 5,680 m³/day = 237 m³/h; at 65 L/m²·h, 3,640 m². With 80 m² modules that is 46 modules; install 48 in four racks of 12 so that one rack can be in backwash or CEB while three filter, each rack designed to carry a third of the flow at 87 L/m²·h for short periods.
• Winter TMP check. Fouled permeability 150 L/m²·h·bar at 20 °C gives 0.43 bar at 65 L/m²·h in summer; at 9 °C viscosity is 1.34 mPa·s, so TMP rises to 0.58 bar, and to 0.78 bar during the three-rack condition, inside a 1.5 bar limit with margin for fouling between CIPs.
• Energy. Feed pumping to about 1.5 bar, backwash and CEB pumps, strainers and drum filter: 0.08 to 0.12 kWh/m³ for the pressurised configuration, against 0.14 to 0.31 kWh/m³ for a submerged plant where blowers are 86 to 93 percent of the load.

Published whole-life figures for a full-scale pressurised UF tertiary plant in Spain give capital of €0.027 to 0.030 per cubic metre treated, with membranes 66 to 69 percent of the investment, operating cost of €0.042 (winter) to 0.077 (summer) per cubic metre dominated by energy, and a total annualised cost of €0.07 to 0.11 per cubic metre. At UK electricity prices and with the lower pressurised energy demand, £0.08 to 0.15 per cubic metre is a reasonable envelope for budgeting, which compares with mains water at £1.5 to 2.5 per cubic metre for the industrial user. Membrane fundamentals and module selection are covered in the ultrafiltration design guide, and the commercial case for reuse in the industrial water reuse article.

UK quality targets and permitting

England has no statutory reuse standard for treated effluent. Schemes are designed to the EU Water Reuse Regulation 2020/741 Class A (E. coli 10 per 100 mL, BOD 10 mg/L, suspended solids 10 mg/L, turbidity 5 NTU, with validation of log reductions) or to the user specification, with BS 8525 applied where the water enters buildings. UF alone meets Class A for solids and turbidity by a wide margin, achieves the bacterial reduction, and is normally followed by UV or chlorination for the residual and virus credit. On permitting, taking effluent for reuse reduces the discharge rather than creating a new abstraction, but the works environmental permit flow conditions, the return of backwash and CEB wastes to the inlet, and any change to the effluent quality reaching the river should be agreed with the Environment Agency in advance; for indirect potable reuse, where UF is the pretreatment to reverse osmosis, the Drinking Water Inspectorate becomes a party as well. Water resources management plans for 2025 to 2050 now include effluent reuse schemes at several large works, and the tertiary UF stage is common to all of them.