Tertiary Cloth and Disc Filtration for Ultra-Low Phosphorus and Solids: Media, Loading, Coagulant and Full-Scale Results

Why filtration follows coagulation

Secondary effluent carries 0.5 to 2 mg/L of total phosphorus, of which about 80 percent is dissolved orthophosphate. Precipitating that fraction with ferric or aluminium salts converts it into fine metal phosphate and hydroxide solids, typically 10 to 50 µm after flocculation, and the effluent total phosphorus then depends almost entirely on how completely those solids are captured. Ten milligrams per litre of effluent solids carries 0.2 to 0.4 mg/L of total phosphorus once the solids are iron phosphate, so a permit at 0.25 mg/L is a solids permit in disguise: the separation stage must deliver 2 to 5 mg/L of solids, and under 2 mg/L for a 0.1 mg/L target.

Cloth filters compete with sand filters, membranes and ballasted flocculation on footprint (a 100 m² sand filter is replaced by a cloth filter tank of about 20 m² plan area), on energy (under 1 W per m³ treated) and on simplicity. Their weakness is that they are a surface filter with little depth: what the floc does not hold, the cloth does not catch, so flocculation upstream matters more than for a depth filter.

Pile cloth versus woven media

Two families of media are sold under the generic label of disc filter, and they behave differently.

• Pile cloth (needle-felt or free-fibre cloth on a woven polyester backing) carries a dense layer of fine fibres that lie flat under forward flow, forming an overlapping matrix with a nominal rating of 10 µm for standard cloth and about 5 µm for microfibre grades. Solids are held within the pile as well as on its surface, which gives some depth-filtration character and tolerance of floc carryover. Discs are fully submerged in an outside-in arrangement and a vacuum shoe draws filtrate back through the cloth during backwash while the disc rotates past it, so filtration never stops.
• Woven polyester mesh panels, with apertures of 10 to 25 µm for tertiary duty and 30 to 100 µm for primary microsieving, operate inside-out. Removal is close to a true sieve cut, with 95 to 99 percent capture above the aperture and little below it, so chemical pretreatment is normally needed for phosphorus. Backwash is by spray jets into a trough while the discs rotate.

Side-by-side pilot work at Merced in California tested a 10 µm pile cloth unit against a woven disc filter on the same secondary effluent; at 2 to 5 NTU influent both met 2 NTU effluent at loadings up to 14.7 m/h without chemicals. When the plant upset lifted influent turbidity to 5 to 10 NTU, the pile cloth held 2 NTU with 5 mg/L of coagulant and 15 to 20 minutes of flocculation at any loading tested, while the woven unit needed roughly four times the coagulant and a loading below 2 gpm/ft² (4.9 m/h). An accidental overdose test at 50 mg/L raised the backwash rate but did not blind the pile cloth. Hence pile cloth dominates UK phosphorus schemes and woven media dominates reuse and microsieving duty.

Hydraulic and solids loading

Two loadings govern the number of discs. Hydraulic loading is flow divided by submerged cloth area; solids loading is the mass of suspended solids (influent solids plus precipitate) arriving per square metre per hour. Published design envelopes:

Practical UK sizing for phosphorus duty sits at 5 to 8 m/h at flow to full treatment with all units in service and 10 to 12 m/h with one unit out, with solids loading checked against 0.1 to 0.15 kg/m²h, beyond which backwash frequency climbs steeply. Cannington, a Wessex Water site, uses a single 30 m² unit for 35 L/s (4.2 m/h) to take 46.5 mg/L of solids down to 10 mg/L. Winsford, a United Utilities trickling filter works of 30,481 population equivalent, uses three 60 m² units for a peak of 28,400 m³/d, which is 6.6 m/h with all three running and 9.9 m/h with one off, against a 1 mg/L total phosphorus consent.

Coagulant dose, mixing and flocculation time

Dose is set by the molar ratio of metal to phosphorus, not by milligrams per litre alone, because the stoichiometric 1:1 ratio only holds while orthophosphate is abundant. As the residual falls below about 0.3 mg/L, each further increment of removal demands disproportionately more metal, since the precipitate is increasingly hydroxide with phosphate adsorbed onto it. Lund pilot work on woven microsieves found 1 to 5 mg Me³⁺/L, equivalent to 5 to 7 mol Me per mol of influent total phosphorus (10 to 14 mol per mol orthophosphate), was required for a consistent 0.1 mg/L, with 0.6 to 2 mg/L of cationic polymer (0.05 to 0.12 mg per mg of influent solids) to build a floc the mesh could hold. For a 0.3 to 0.5 mg/L target with pile cloth, 1 to 9 mg Fe/L sufficed in the Cranfield demonstration, with no polymer under steady conditions.

Mixing conditions matter as much as dose. Dispersion of the coagulant needs a velocity gradient of 400 to 1,000 s^-1 for a few seconds, which an in-line static mixer or an orifice plate ahead of the dosing point provides; raising it to 2,000 s^-1 brought no further benefit. Coagulation then needs about 1 minute and flocculation about 4 minutes at the minimum, which is what the Arvidstorp full-scale plant in Sweden was built to. For pile cloth fed with ferric the Cranfield rig used a 330 L rapid mix at 290 rpm and a 1,650 L flocculator slowed from 130 to 25 rpm, giving 7 to 27 minutes of flocculation across the flow range, and the authors ranked flocculation optimisation as the key to robust performance. At Merced the pile cloth needed 15 to 20 minutes of flocculation to hold 2 NTU at 5 mg/L dose; 10 minutes was not enough. A good design therefore gives 10 to 20 minutes of gentle flocculation at peak flow rather than relying on pipe flocculation alone.

Two-point dosing, with the bulk of the iron into the primary or secondary stage and a trim dose of 2 to 5 mg Fe/L ahead of the tertiary filter, cuts the tertiary precipitate load and keeps the iron residual low; UK iron consents are commonly 3 to 4 mg/L as a 95 percentile. Over-dosing shows up as iron in the effluent before it shows up as phosphorus, so an online iron or turbidity signal on the filtrate is a useful guard. Chemical dosing system design covers the storage and control side.

Backwash, headloss and cleaning

A pile cloth filter runs with the discs stationary and fully submerged. Solids collect on and in the pile, the tank level rises above the fixed outlet weir, and at a set differential (typically 100 to 250 mm) the discs rotate, the suction shoes draw filtrate back through the cloth, and the wash water is pumped to the head of the works. Backwash volume is normally 2 to 5 percent of throughput at design solids loading, rising sharply if that loading is exceeded. Operating headloss across the unit at all flows is under 250 mm, which lets the filter drop into an existing hydraulic profile between humus tanks and outfall without pumping in many cases.

Cloth life is quoted at up to 8 years. Periodic chemical cleaning recovers capacity lost to biofilm and iron fouling that the backwash cannot remove: woven media at Rya in Gothenburg is washed with 10 to 15 percent hydrochloric acid and 4 percent hypochlorite about every six weeks, and panel tests there showed filtration velocity falling up to 49 percent between cleans. Uneven flow split between parallel units causes uneven fouling, so inlet weirs or manifolds must divide flow within a few percent; Arvidstorp recorded one unit losing half its capacity in 90 days while its neighbour was barely loaded, purely because of the split.

What the trials show about achievable phosphorus

Demonstration-scale work for Severn Trent at Cranfield compared pile cloth filtration, magnetite-ballasted flocculation and ultrafiltration, all with ferric chloride, against a full-scale continuously backwashed sand filter, with an effluent target of 0.3 mg P/L. Cloth filtration gave a median effluent orthophosphate of 0.18 mg/L, 0.10 to 0.24 mg/L up to the 90th percentile and a maximum of 0.56 mg/L, with spot-sample total phosphorus averaging 0.30 ± 0.27 mg/L; ballasted flocculation and the sand filter gave 90th percentiles of 0.07 and 0.03 mg/L. On a robustness index (90th to 50th percentile scaled to the target, where below 1 is good), cloth filtration was robust for targets of 0.3 to 0.5 mg/L but not for 0.1 mg/L, and a conservative operational target of 0.21 mg/L orthophosphate left it with almost no margin.

Full-scale 12-month trials under the national phosphorus programme (CIP2) found that a continuously backwashed upflow sand filter dosed at 10 to 14 mg Fe/L reached 0.22 mg P/L, while a high-rate compressible media filter dosed at 4 mg Fe/L managed only 1.7 mg/L because of insufficient reaction time and iron carryover. Where the target is 0.1 mg/L the evidence favours ballasted flocculation or a filter with real depth; see the companion guide on ballasted flocculation clarifiers.

In UK regulatory terms the Technically Achievable Limit for phosphorus is 0.25 mg/L as an annual average, with some sites now permitted at 0.1 mg/L, assessed on spot samples over the year. An annual average forgives short excursions but not seasonal drift, so dose control, flocculation time at winter temperatures and solids loading during storm return flows are the three things to prove in the first winter. For other routes to the same permit see tertiary treatment equipment.

How to size a tertiary cloth filter for a phosphorus consent

1. Fix the solids balance. Take influent solids at the 95 percentile, add the chemical precipitate from the planned dose, and derive the solids load to the filter at flow to full treatment.
2. Choose the dose by molar ratio. About 2 mol Fe per mol influent P for a 0.5 mg/L target, 5 to 7 for 0.1 mg/L, split between an upstream point and a tertiary trim dose.
3. Size on both loadings. 5 to 8 m/h and under 0.15 kg/m2h with all units running; check the n minus 1 case at 10 to 12 m/h.
4. Provide real flocculation. Static mixer for dispersion, then 10 to 20 minutes of gentle flocculation at peak flow with a variable speed mixer.
5. Close the hydraulic profile. Allow 250 mm across the filter plus the inlet split structure, and confirm free discharge at the outfall under flood level.
6. Return the backwash. Route 2 to 5 percent of flow back to the inlet works or primary tanks and include it in the works load.