Anaerobic Membrane Bioreactors: Energy Balance, Dissolved Methane and Honest Design Limits

What an AnMBR is and why the combination matters

Conventional anaerobic reactors — UASB, EGSB, anaerobic filters — rely on the biomass settling or granulating to stay in the reactor. That works well on soluble, warm, well-balanced wastewaters and poorly on everything else: dilute streams, particulate-rich feeds, variable loads, and any water whose chemistry discourages granulation.

Replacing gravity separation with a membrane removes that dependence entirely. The consequences are structural:

• Complete biomass retention. SRT is set by wasting alone and can be 50–100 days or more, independent of hydraulic retention time.
• Slow growers are retained. Methanogens with doubling times of days to weeks are never washed out, so the process tolerates dilute and cold feeds that would fail a settling-dependent reactor.
• Absolute solids barrier. Permeate is free of suspended solids regardless of sludge settleability, so effluent quality is decoupled from biological condition.
• Particulates are hydrolysed rather than lost. Long SRT allows slowly biodegradable particulate matter time to be converted rather than passing through.

The trade-off is that the membrane must be kept clean in an environment with no oxygen, high solids and abundant soluble microbial products — a considerably harder fouling problem than in an aerobic MBR.

The energy case, worked properly

The energy argument is the reason AnMBR exists, so it deserves an explicit balance rather than an assertion.

Per kilogram of COD removed:

• Gross energy = 0.35 × 9.94 = 3.48 kWh thermal. In practice 5–10% of COD goes to biomass synthesis, so take roughly 3.2 kWh.
• Through a CHP engine at 35% electrical efficiency: 1.12 kWh electrical plus useful heat.
• Aerobic treatment of the same kilogram requires roughly 1.0 kWh of aeration energy (0.7–1.2 depending on transfer efficiency and alpha).
• Net swing ≈ 2.1 kWh per kg COD.

For an industrial effluent of 500 m³/d at 5,000 mg/L COD with 90% removal, that is 2,250 kg COD/d removed: about 2,520 kWh/d of electricity generated instead of roughly 2,250 kWh/d consumed — an annual swing of order 1,700 MWh.

This is the single most important selection rule for AnMBR: the energy case scales with COD concentration, and the parasitic load scales with volume. Strong, warm industrial effluents are the natural application; dilute municipal sewage at temperate temperatures is where the case becomes marginal.

Dissolved methane: the problem that decides the design

Methane obeys Henry’s law like any other gas, and the permeate leaves the reactor saturated with it. That dissolved methane is lost energy and, worse, a potent greenhouse gas emission.

Worked value. At 35 °C in a biogas of 70% methane, p_CH4 = 0.7 atm, so C_sat = 1.0×10⁻³ × 0.7 = 7.0×10⁻⁴ mol/L × 16 g/mol = 11.2 mg/L. In practice permeate is often supersaturated relative to this equilibrium, and values of 10–20 mg/L are commonly reported.

For the 500 m³/d example at 12 mg/L:

• Dissolved methane lost = 500 × 12 = 6,000 g/d = 6.0 kg CH₄/d.
• Total methane produced = 2,250 kg COD × 0.35 m³/kg × 0.657 kg/m³ ≈ 517 kg/d, so the loss is about 1.2% of production — trivial as energy.
• As emissions it is not trivial: 6.0 kg/d × 27 (GWP₁₀₀) = 162 kg CO₂e/d = 59 tCO₂e/yr, materially eroding the carbon benefit of the energy recovered.

Recovery options are vacuum degassing, membrane contactors (hollow-fibre degassers, the most efficient but the most capital-intensive), and downstream aerated polishing which strips the methane to atmosphere — solving the dissolved oxygen demand problem while making the emissions problem worse. Any AnMBR carbon claim that does not account for dissolved methane should be treated as incomplete.

Fouling without oxygen

Anaerobic membrane fouling is more severe than aerobic, for identifiable reasons.

• No air scour. Biogas sparging must do the same job with a gas that must be recirculated, compressed and contained — which is where most of the parasitic energy goes.
• More soluble microbial products and EPS. Anaerobic biomass produces gel-forming material that fouls readily and reversibly, then irreversibly.
• Inorganic precipitation. Struvite and calcium carbonate precipitate at the membrane surface where local pH and supersaturation are highest — a scaling problem on top of a biofouling problem.
• Fine, poorly filterable particles. Long SRT accumulates inert fines that build a compressible cake with high specific resistance.

The consequence is that sustainable flux is low: typically 5–15 L/m²·h, against 15–30 for an aerobic MBR. Membrane area, and hence capital cost, is correspondingly higher for the same flow — another reason the economics favour concentrated feeds where the flow is modest for the load.

The general fouling framework and cleaning chemistry are covered in our guide to membrane fouling and CIP.

Sulphate, temperature and the other constraints

Three feed characteristics decide whether an AnMBR will behave as designed.

• Sulphate. Sulphate-reducing bacteria compete with methanogens for hydrogen and acetate and generally win thermodynamically. Every gram of sulphate reduced consumes about 0.67 g of COD that would otherwise have become methane, and produces hydrogen sulphide — toxic, corrosive and a gas-cleaning cost. As a rule of thumb, a COD:SO₄ ratio below about 10 warrants serious attention, and below about 5 the methane yield is materially compromised. See our hydrogen sulphide guide for the downstream consequences.
• Temperature. Mesophilic operation at 35 °C gives the best kinetics but requires heating, which on a dilute stream can consume more energy than the methane returns. Psychrophilic operation at 15–25 °C avoids the heating load but halves or quarters the rate, requiring a longer SRT and a larger reactor — usually acceptable because the membrane guarantees retention regardless.
• Nitrogen and phosphorus. Anaerobic treatment does not remove them. Organic nitrogen is mineralised to ammonium, so permeate ammonia is typically higher than the influent total ammoniacal nitrogen. Any consent with a nitrogen limit requires downstream treatment — conventionally nitrification and denitrification, or more elegantly anammox, which suits a carbon-depleted, ammonium-rich permeate very well.

That last point is the most commonly underestimated. An AnMBR is a COD-removal and energy-recovery process, not a complete treatment plant. It must be designed as the front end of a train.

Typical design parameters

Alkalinity deserves a separate check. Anaerobic processes require adequate buffering to absorb volatile fatty acid production; a total alkalinity of 2,000–4,000 mg/L as CaCO₃ and a VFA-to-alkalinity ratio held below about 0.3–0.4 is the conventional stability criterion. Feeds low in alkalinity need supplementation, and that chemical cost belongs in the operating comparison.

Where AnMBR makes sense — and where it does not

The honest summary is that AnMBR is an excellent industrial technology whose municipal case remains conditional on solving dissolved methane recovery economically. For a UK food or beverage site with a strong effluent, a heat demand and a rising trade effluent bill, it is a genuinely compelling option — and, as always, a treatability trial precedes any commitment, because biodegradability, sulphate and filterability are all feed-specific. That evidence-first approach is standard practice in anaerobic and membrane process selection.