Ballasted Flocculation Clarifier Design: Microsand and Magnetite Systems for Phosphorus and Storm Flows

How ballast changes the settling problem

A metal hydroxide floc has a density only a few percent above that of water and a settling velocity of 1 to 3 m/h at best, which is why clarifiers for chemically treated water are sized at 1 to 2 m/h surface loading and why lamella plates or dissolved air flotation are used to shrink them. If the floc is built around a grain of silica sand of 100 to 150 µm (density 2,650 kg/m³) or of magnetite (about 5,000 kg/m³), the composite settles at tens of metres per hour, and the clarifier can be loaded at 40 to 150 m/h with lamella tubes to polish the last fraction. Published footprints are five times smaller than a lamella clarifier or DAF and up to twenty times smaller than a conventional tank for the same flow.

Ballast also stabilises the process. With several grams per litre of sand or magnetite in circulation, the ballast mass dwarfs the incoming solids, so a step change in influent turbidity or flow barely alters floc density; Seattle pilot work showed effluent quality recovering within one to two residence times of a disturbance, and the systems can go from idle to full performance within about 15 minutes, which suits storm duty. Shear is the price: the polymer must bind floc to ballast firmly enough to survive the mixers, so polymer selection and dose matter more than in a conventional clarifier.

Process stages and residence times

A microsand system has four stages in series, each a mixed tank, followed by the settler. Coagulant (ferric chloride, ferric sulphate, alum or polyaluminium chloride) is dosed into the feed, often through a static mixer, and the coagulation tank at 1 to 2 minutes forms pin floc. Microsand and polymer are added in the injection tank, 1 to 2 minutes at high shear, where polymer bridging attaches floc to sand. A maturation tank at 3 to 6 minutes with gentler mixing grows the ballasted floc. Settling takes place in a tank with a scraper and lamella tubes, from whose underflow the sand and sludge mixture is pumped to hydrocyclones. Some designs replace the separate injection and maturation tanks with a draft-tube turbo mixer that recirculates the ballasted floc internally.

Bench jar tests mirror the plant: coagulant and sand at 300 rpm for two minutes, polymer at 200 rpm for 45 seconds, then two minutes of settling. Wet weather pilots in Tennessee held settled turbidity below 2 NTU at 73 to 98 m/h with 65 to 130 mg/L of ferric chloride and 2.5 to 4.5 mg/L of polymer on raw sewage blended with activated sludge; tertiary duty on secondary effluent needs an order of magnitude less coagulant.

Coagulant, polymer and ballast balance

For tertiary phosphorus the coagulant is set by the metal to phosphorus molar ratio, as for any chemical precipitation, but the ballasted clarifier can afford to run at the high end of the range because its sludge is captured so completely. Milwaukee pilot work on secondary effluent held total phosphorus at 0.02 to 0.05 mg/L with 7 to 12 mg Fe/L of ferric chloride at rise rates of 34 to 41 m/h; the Cranfield demonstration of a magnetite system for Severn Trent used 3 to 8 mg/L of ferric chloride and 0.5 to 1 mg/L of polyacrylamide with magnetite at 4 g/L in circulation and 25 percent of settled sludge recycled, and gave a median effluent orthophosphate below the 0.05 mg/L detection limit with a 90th percentile of 0.07 mg/L. Anionic polymers of low to moderate charge (minus 10 to minus 15 percent) have outperformed cationic products on most waters, at 0.5 to 1 mg/L for tertiary duty and 2.5 to 4.5 mg/L for raw wastewater.

Sand size and concentration are design variables in their own right. A drinking water plant in Arizona that struggled with settled turbidity found that changing from a coarser to a finer sand grade cut settled turbidity from 1.5 to 0.6 NTU at the same polymer dose and halved the polymer demand, because the finer grade offers more surface per gram for the same ballast mass; performance also improved as sand concentration in the maturation tank rose. The trade-off is sand loss: finer grains are carried over to the effluent and lost in the hydrocyclone overflow more readily. Make-up rates of 1 to 3 g of sand per cubic metre treated are typical, and sand carryover into a downstream UV or filter stage must be checked.

Ballast recovery: hydrocyclones and magnetic drums

Recovery is what makes ballast affordable. In a microsand system the settler underflow, a mixture of sludge and sand, is pumped continuously to hydrocyclones. Centrifugal separation sends the dense sand to the underflow, which is returned to the injection tank, and the light sludge to the overflow, which goes to sludge treatment. On the Seattle pilot the cyclone feed was 7 percent of plant flow (22 gpm against 310) and the underflow 20 percent of the cyclone feed, so about 1.4 percent of plant flow went back as sand slurry and about 5.7 percent left as sludge. Handbook guidance stresses that the cyclone must receive a dilute sludge to classify cleanly; the sand can then make up as much as 95 percent of the flocculated solids mass in circulation. Sludge leaves at 0.2 to 0.5 percent dry solids and usually needs thickening, so the sludge line is sized on volume as much as mass.

Magnetite systems replace the cyclone with a magnetic drum: the sludge recycle stream is passed over a rotating drum with permanent magnets, the magnetite is stripped from the floc by shear mixing first and then recovered on the drum at better than 99 percent, and the non-magnetic sludge goes to waste. Magnetite is denser than sand, so lower circulating concentrations are needed, and the reported clarifier can be up to 90 percent smaller than a conventional tank. Either way, ballast inventory and recovery efficiency belong in the performance guarantee, because a steady loss of ballast shows up first as a slow decline in settled quality and only later as a stores invoice.

UK phosphorus applications

Under AMP7 the ballasted clarifier became the UK answer to permits below the 0.25 mg/L technically achievable limit. Thames Water installed two packaged microsand units at East Shefford, discharging to the River Lambourn chalk stream, against a 0.25 mg/L annual average permit and a 0.1 mg/L stretch target, having seen two earlier UK units achieve 0.08 and 0.10 mg/L; the works alternates duty between the units at low flow so that both stay ready for the sudden rises its infiltration-prone catchment produces. Southern Water adopted the same technology at Hailsham South, a 30,162 population works with a maximum flow to treatment of 538 L/s, to push beyond the 0.25 mg/L limit to 0.1 mg/L, after a scale-model pilot on site, with the vendor quoting rise rates of 80 to 150 m/h and up to 200 m/h for that turbo-mix configuration. Severn Trent chose a magnetite system at Cannock, where the average flow is 226 L/s and flow to full treatment 486 L/s, to meet a 0.3 mg/L permit that applied from December 2021; reported effluent total phosphorus is frequently below 0.1 mg/L with iron below 1 mg/L against a 3 mg/L consent.

Demonstration-scale comparison at Cranfield for Severn Trent put the magnetite system against pile cloth filtration, ultrafiltration and a full-scale sand filter at a 0.3 mg P/L target. Ballasted flocculation was the most robust of the three new technologies, with 90th percentile orthophosphate of 0.07 mg/L and a maximum of 0.84 mg/L over steady state, storm and diurnal perturbations, and spot-sample total phosphorus averaging 0.05 mg/L. Cloth filtration was judged suitable for 0.3 to 0.5 mg/L targets but not 0.1 mg/L. Where the permit is 0.1 mg/L, ballasted flocculation therefore carries the evidence, and where it is 0.5 mg/L a cloth filter at a fraction of the chemical and sludge cost is usually enough; the companion guide on tertiary cloth filtration sets out that case.

Two UK specifics deserve attention. Polymer residual in the effluent is a toxicity concern on protected watercourses; Southern Water commissioned an aquatic toxicology assessment for the Hailsham invertebrates before consenting to continuous polymer dosing, and the Environment Agency may ask the same of any site discharging to a Special Area of Conservation. Iron residual is the other: at 10 mg Fe/L dosed, a 3 mg/L iron consent leaves little margin if the settler is overloaded, so an online iron or turbidity signal on the clarified water with dose trimming is prudent. Chemical dosing design covers the storage, bund and control requirements.

Storm and combined sewer overflow duty

The same equipment, run at the top of its rise rate range, treats the storm flows that would otherwise bypass biological treatment. Vendor performance data for wet weather duty at 147 m/h give total suspended solids below 20 mg/L, BOD below 30 mg/L and phosphorus of 0.5 to 1.5 mg/L from screened raw sewage, which is better than a storm tank and comparable with primary settlement at twenty times the loading. With an aerated contact tank ahead of the clarifier, in which return activated sludge adsorbs soluble BOD for 10 to 25 minutes before ballasted settling, total BOD removal above 85 percent has been demonstrated at contact tank solids above 1,000 mg/L. For a UK works facing a storm overflow spill reduction target, a ballasted clarifier that serves as tertiary phosphorus plant in dry weather and storm treatment in wet weather is an attractive dual-use asset, provided dosing and sludge handling are sized for the storm case. For the wider choice between settlement, flotation and ballasted systems see our DAF and clarification guidance and the overview of tertiary treatment equipment.