The Cost of Phosphorus Removal to Low Consents: Capex, Opex and Whole-Life Cost per Kilogram

Why cost is convex in the permit limit

Removing phosphorus from 8 mg/L to 1 mg/L takes out 87 percent of the load with simple co-precipitation: ferric or aluminium salts dosed at about 1.5 mol metal per mol phosphorus, settled in the existing clarifier. Going from 1 to 0.5 mg/L removes a further 6 percent of the load but needs a metal to phosphorus ratio of 2 to 3, because residual orthophosphate falls towards the solubility and adsorption limits of the metal hydroxide floc. Below 0.3 mg/L the residual is dominated by particulate and organically bound phosphorus that no dose will precipitate, so a solids barrier (cloth, sand or membrane filtration) becomes mandatory and the metal ratio climbs to 4 to 6 for the last fractions of soluble P. Each step therefore costs more in chemicals, more in sludge (roughly 2.5 kg dry solids per kg iron dosed) and, from about 0.5 mg/L down, a whole new tertiary stage with its own civil, mechanical and electrical cost. Plotting annualised cost against the permit gives a curve that is nearly flat from 2 to 1 mg/L and nearly vertical below 0.25 mg/L.

International cost data

Jiang and co-workers’ US study, reproduced in the Montana DEQ nutrient cost review, gives total annual economic cost (capital recovery at 8.72 percent over 20 years plus operation) for upgrading a conventional activated sludge plant with alum dosing, at 2004 prices:

Converting the 10 MGD column to marginal cost per kilogram is instructive. Tightening from 1.0 to 0.5 mg/L removes an extra 0.5 g/m³ × 37,850 m³/d = 6,900 kg P/yr for $1.37 M/yr, about $200/kg. From 0.5 to 0.13 mg/L the extra 5,100 kg/yr costs $2.0 M/yr, roughly $390/kg. From 0.13 to 0.05 mg/L only 1,100 kg/yr more is removed for $1.84 M/yr, about $1,670/kg. Average cost per kilogram across the whole plant stays below $100/kg because the first milligrams are cheap, which is why average figures mislead when a permit is being negotiated: it is the marginal cost that the last 0.1 mg/L has to justify.

Bashar and co-workers (2018) modelled six 5 MGD configurations in BioWin and found unit costs of $42 to $61 per lb of P removed ($93 to $134/kg) across biological, membrane and tertiary options; the MUCT process with chemical trim and tertiary reactive media filtration reached 0.05 mg/L at $44/lb ($97/kg), marginally dearer than IFAS-EBPR at 0.8 mg/L, because the extra capital was spread over a near-total removal. Membrane bioreactors were the least cost-effective at $61/lb, their effluent no better than 1 mg/L without chemical trim. Eawag’s 2008 review for the state of Hessen found that simply optimising simultaneous precipitation to 0.5 mg/L cost 12 to 27 euro per additional kilogram removed, while two-layer filtration to reach 0.1 to 0.2 mg/L cost 84 to 133 euro/kg and ten times the capital and operating outlay of conventional dosing. Wisconsin’s state-wide screening of 195 plants found an average capital cost of $3.75 M per plant for the group needing to reach below 0.1 mg/L.

Ofwat PR24 enhancement cost models

Ofwat does not publish a unit rate per site; it sets allowances with four scheme-level econometric models (two on PR19 historical outturn, two on PR24 forecasts, each weighted 25 percent) whose explanatory variables define what the regulator believes drives cost:

• Population equivalent served: scale, with strong economies; average totex per PE in the PR24 programme is about £400, double the PR19 figure of £200, because the average scheme is a smaller works (17,000 to 20,000 PE) receiving a tighter permit.
• Enhanced permit and its square: the continuous non-linear rise in cost as the permit tightens; the weighted average enhanced permit in PR24 is 0.53 to 0.58 mg/L.
• Historical permit: whether dosing already exists, so that a move from 2 to 0.25 mg/L is costed as an upgrade rather than a new build.
• Technically achievable limit dummy: a step change for permits at or below 0.25 mg/L, the level the UKWIR Chemical Investigations Programme established as the practical floor for full-scale plant, capturing the discrete cost of a tertiary stage.

Schemes with permits below 0.25 mg/L or with a biological treatment component were treated as engineering outliers and allowed 75 percent of the gap between request and model, and the models applied an overall industry efficiency challenge of 15 percent. Company submissions illustrate the resulting envelope: Dŵr Cymru’s PR24 case covers 149 phosphorus sites for £153 M of enhancement totex at 2022/23 prices, including 92 Welsh sites given 5 mg/L backstop permits for the first time, with named examples of a works moving from 2 to 0.25 mg/L needing a new tertiary stage despite existing dosing, and another moving from 1 to 0.7 mg/L needing only dosing optimisation. Severn Trent’s WINEP submission built unit-rate models by works size and consent level and reported cloth media filters as its standard tertiary solids step, with reactive media and apatite beds as lower-energy alternatives at small sites.

For a designer the practical reading is this: anything at or below 0.25 mg/L will be scrutinised against a model that already assumes a tertiary stage, so the cost case must show the filter sized on the right solids load, the chemical ratio justified by jar testing, and the sludge route costed. Regulatory background on the permit side is in the nutrient neutrality and phosphorus consents guide.

Technology options against the permit

Biological phosphorus removal shifts cost from opex to capex and sludge age; it reaches 0.5 to 1 mg/L reliably on strong sewage with a readily biodegradable COD to P ratio above about 15, but UK works receiving dilute, cold, nitrifying flows rarely sustain it below 0.5 mg/L without chemical trim, and the Ofwat models treat biological schemes as outliers for exactly that reason. Side-stream struvite recovery reduces the recycle load and the metal demand but, as Bashar’s mass balance showed, leaves the final effluent largely unchanged, so it is a sludge-line economy rather than a consent strategy; see the struvite recovery article. For the tertiary technologies themselves, the ultra-low phosphorus tertiary removal guide compares cloth, sand, ballasted and membrane options.

Worked UK example: 20,000 PE, 1.0 to 0.25 mg/L

A works treating 4,000 m³/d dry weather flow with existing simultaneous ferric dosing to 1.0 mg/L must meet 0.25 mg/L annual mean. Design for 0.2 mg/L to give compliance margin.

• Extra phosphorus removed: 0.8 g/m³ × 4,000 m³/d = 3.2 kg/d = 1,170 kg/yr.
• Capex: tertiary cloth disc filter for 3 × DWF with flocculation chamber, second dosing point, sludge thickening uplift, MEICA and civils: £5.0 M, consistent with Ofwat’s PR24 average of about £400 per PE for a scheme of this size and permit. Annuity at 3.5 percent over 25 years (factor 0.0607): £303,000/yr.
• Chemical: the last 0.8 mg/L needs Fe:P near 4:1 molar, 0.8/31 × 4 × 55.85 = 5.8 mg Fe/L, 23 kg Fe/d, about 170 kg/d of 40 percent ferric chloride, 62 t/yr at £260/t: £16,000/yr.
• Sludge: 2.5 kg DS per kg Fe plus the captured solids, about 25 t DS/yr at £320/t DS to cake and recycle: £8,000/yr.
• Power: filter drives, backwash and interstage pumping average 12 kW, 105 MWh/yr at £0.20/kWh: £21,000/yr.
• Maintenance and operator time: 2 percent of capex plus 0.25 FTE: £115,000/yr.

Had the same works been asked for 0.5 mg/L, a second dosing point, a polymer-aided clarifier upgrade and better dose control at £0.8 M would remove 0.5 g/m³ × 4,000 = 730 kg/yr for an annualised £70,000 plus £20,000 opex: about £125/kg. Three times the cost per kilogram, then, for the step from 0.5 to 0.25 mg/L, and the step to 0.1 mg/L with ballasted flocculation would roughly triple it again. Those ratios match the US and Swiss curves above, which is a useful cross-check on any UK estimate.

Pitfalls in cost estimates

• Costing the annual mean as a daily limit. UK phosphorus permits are annual means. A plant designed to hit 0.25 mg/L every day is oversized; one designed on the mean with a 0.05 mg/L margin and good dose control is cheaper and still compliant.
• Ignoring the sludge chain. Metal hydroxide sludge dewaters less well and may breach the iron limits of agricultural recycling; digester capacity, cake tonnage and haulage belong in the opex.
• Forgetting return liquors. Backwash and thickener liquors recycle solids-bound phosphorus and can add 10 to 20 percent to the load on the dosing stage.
• Comparing capital across decades without indexation. Ofwat’s own PR24 unit costs per PE doubled against PR19 for reasons of scope, not only inflation.
• Omitting metal residuals. Iron and aluminium in the effluent carry their own limits, and higher Me:P ratios at low permits push them up.

Guidance on the chemistry and dose control behind these numbers is in the chemical phosphorus removal article.