Membrane Bioreactor Design: Flux, MLSS and Aeration Energy

What is a membrane bioreactor and how is it designed?

An MBR is a suspended-growth biological process in which a micro- or ultrafiltration membrane, rather than gravity settling in a clarifier, separates the treated water from the biomass. Because the membrane retains all suspended solids and flocs down to roughly 0.04–0.4 µm, the biological reactor is completely decoupled from the sludge-settling behaviour that constrains a conventional activated-sludge plant. The designer is no longer held hostage to the sludge volume index (SVI): filamentous bulking, pin floc and rising sludge do not carry solids into the effluent because there is no overflow weir to escape from. That single change — separation by pore size instead of by density — is what lets an MBR run at mixed-liquor suspended solids (MLSS) of 8–12 g/L, roughly three to four times the 2.5–4 g/L typical of conventional plants.

Designing an MBR therefore proceeds on two coupled tracks that must be reconciled. The biological design fixes the reactor volume, the solids retention time (SRT), the aeration required to oxidise carbon and ammonia, and any anoxic or anaerobic zones for nutrient removal — the same mass-balance logic used for activated-sludge process design, but at elevated biomass concentration. The membrane design fixes the installed membrane area from the peak flow and the sustainable flux, and sizes the air scour and cleaning regime that keep the membranes permeable. The two tracks meet at the aeration system, because the high MLSS that makes the process compact also degrades oxygen transfer, and the air used to scour the membranes is usually the single largest energy line item in the whole works.

The reward for that complexity is effluent quality. An MBR routinely produces permeate below 1–2 mg/L BOD and turbidity under 0.2 NTU, essentially free of suspended solids and with a large log-reduction of bacteria and protozoa. That reuse-grade quality — suitable for feeding reverse osmosis or for direct non-potable reuse — is the reason MBRs are chosen despite their higher energy and capital cost. Whether that trade-off is worth making against a coupled fixed-film alternative is the subject of our companion piece on the choice between MBR and MBBR processes.

Why design at high MLSS and long SRT?

The membrane lets the designer choose the biomass concentration almost freely, and the usual choice is a high one: 8–12 g/L MLSS, occasionally up to 15 g/L. The attraction is a compact biological reactor. For a given organic load the required reactor volume falls roughly in inverse proportion to MLSS, because it is the mass of active biomass — concentration multiplied by volume — that determines the treatment capacity, not the volume alone. Tripling the MLSS from 3.5 to 10.5 g/L cuts the biological tank volume to about a third for the same food-to-microorganism ratio, and that is the headline footprint advantage of the technology.

Complete solids retention makes the SRT a clean design variable, set purely by how much sludge is wasted. Long SRTs of 15–30 days are normal, and they deliver three things at once: reliable nitrification even at low temperature, because slow-growing autotrophs are never washed out; a low observed sludge yield, because endogenous respiration consumes a larger fraction of the biomass; and stable operation through load swings. There are limits. As SRT lengthens the MLSS climbs unless volume is added, and both viscosity and the accumulation of inert and colloidal material rise, which suppresses membrane permeability. Beyond about 12–15 g/L the aeration penalty and the fouling penalty outrun the footprint benefit, so most designs settle in the 8–12 g/L band as the practical optimum rather than pushing MLSS as high as the membrane will physically tolerate.

How do you size the membrane area from flux?

Membrane sizing rests on a single relationship between the flow to be filtered, the flux the membrane can sustain, and the area installed. Flux is the permeate flow per unit membrane area, conventionally quoted in litres per square metre per hour, abbreviated LMH (L/m²·h). Rearranged, the required area is simply the design flow divided by the design flux.

The subtlety is which flux to use, and here the distinction between gross and net flux is central. Membranes do not filter continuously. Immersed systems periodically stop permeating to relax (permeation off, air scour on) or run a backwash (permeate pushed back through the membrane) to lift the fouling cake. Time spent relaxing or backwashing produces no net water, so the net flux — the value that actually sizes the plant — is always lower than the instantaneous gross flux the membrane runs at while filtering.

The governing case is nearly always the peak flux, not the average. A membrane plant must pass the peak hydraulic flow — storm flow, diurnal morning peak — through a fixed area, so the peak flux is what determines whether the cake stays reversible. Sustainable-flux thinking underpins the choice of design value: the sustainable flux is the flux below which the transmembrane pressure (TMP) rises only slowly and predictably, so cleaning intervals stay long, as opposed to the critical flux above which fouling accelerates and TMP runs away. Prudent design sets the average-flow flux comfortably below the sustainable value and confirms the peak flux can be held for the required duration by relaxation and air scour alone, with chemical cleaning reserved as recovery, not routine.

Worked example — membrane area. Take an average flow of 100 m³/h with a peak of 200 m³/h. Choose a design net flux of 20 LMH at average flow. Average flow is 100,000 L/h, so A = 100,000 / 20 = 5,000 m². Now check the peak: at 200,000 L/h across 5,000 m² the net flux is 40 LMH. If the membrane’s permitted sustained peak is 40 LMH for up to four hours, the area is adequate; if the peak flux ceiling is 35 LMH, the area must rise to 200,000 / 35 = 5,715 m². The peak governs, and the number of membrane modules (each of known area) is then rounded up from the governing area with an allowance for one cassette out of service for cleaning.

Why does membrane air scour dominate energy demand?

The defining energy cost of an MBR is not the biological aeration but the air blown along the membrane surface to keep it clean. Coarse bubbles rising past the membrane sheets or fibres create shear and two-phase turbulence that scours the fouling cake and sweeps solids back into the bulk. This membrane air scour runs whenever the membranes are in service, and it is quantified by the specific aeration demand, expressed two ways.

SAD_p is the more revealing figure because it ties scour air directly to water produced, and it is where a decade of membrane development has concentrated: cyclic (intermittent) aeration that alternates air between banks of modules, and coarser but less frequent bursts, have pushed SAD_p down from around 40–60 in early plants to nearer 10–20 today. Because the scour blowers may account for a third to a half of total plant energy, a modest reduction in SAD_p moves the whole-life cost of the works more than almost any other design lever.

Worked example — scour energy. For the 5,000 m² plant above, run cyclic scour at an effective SAD_m of 0.3 Nm³/m²·h. Air flow = 0.3 × 5,000 = 1,500 Nm³/h. Blower power for near-atmospheric membrane submergence is roughly P (kW) ≈ Q_air (Nm³/h) × specific energy; taking about 0.02 kW per Nm³/h at 0.5 bar discharge gives P ≈ 1,500 × 0.02 = 30 kW. Against 100 m³/h of permeate that is 0.30 kWh/m³ for scour alone — before biological aeration — which is why total MBR specific energy sits around 0.6–1.2 kWh/m³, well above a conventional plant. Cross-checking, SAD_p = 1,500 / 100 = 15, comfortably in the modern range.

How does high MLSS affect biological aeration?

The biological aeration that supplies oxygen for carbon and ammonia oxidation is a separate system from the membrane scour, and high MLSS makes it harder to deliver. Oxygen dissolves from fine bubbles into the mixed liquor, and the efficiency of that transfer is captured by the alpha factor, the ratio of the oxygen mass-transfer coefficient in the process liquor to that in clean water. In a conventional plant alpha is around 0.5–0.65; in an MBR at 10–12 g/L it falls to roughly 0.4–0.5 and can drop below 0.3 at the highest concentrations. The mixed liquor becomes more viscous and non-Newtonian as MLSS rises, bubbles coalesce, and the surfactant-like extracellular material suppresses interfacial transfer — all of which cut the true oxygen delivered per unit of air.

The design consequence is that the installed fine-bubble aeration must be substantially oversized relative to the theoretical oxygen demand, because each cubic metre of air delivers less oxygen than it would in clean water. Ignoring the alpha decline is a classic under-design that leaves a plant unable to nitrify at peak load.

Worked example — alpha-corrected air. Suppose the process oxygen demand (AOTR) is 120 kg O₂/h. A diffuser system rated at a standard oxygen transfer efficiency giving SOTR is derated by the field conditions. If the combined correction — alpha 0.45, the DO-saturation term about 0.75, temperature and fouling factors — multiplies out to roughly 0.30, then the required clean-water SOTR = AOTR / 0.30 = 120 / 0.30 = 400 kg O₂/h. In other words the aeration equipment must be specified for about 3.3 times the actual oxygen demand. At a standard aeration efficiency of 4 kg O₂ per kWh, the blower draws 400 / 4 = 100 kW for process aeration — comparable to, and often larger than, the membrane scour load, and a direct penalty of running at high MLSS.

Flat-sheet vs hollow-fibre and immersed vs sidestream?

Two configuration choices define the physical MBR. The first is membrane geometry — flat-sheet panels or hollow-fibre bundles — and the second is placement, either immersed directly in the bioreactor or in an external sidestream loop pumped under pressure. Immersed vacuum-driven systems dominate municipal and most industrial work because they operate at low TMP and low energy; sidestream pressurised systems survive in high-strength, high-temperature or otherwise hostile industrial duties where robust cleaning and cross-flow control outweigh the extra pumping energy.

Within immersed systems the flat-sheet versus hollow-fibre decision is a genuine engineering trade-off rather than a clear winner. Flat-sheet panels are mechanically robust, tolerate higher MLSS and abrasive solids, and are hard to irreversibly clog, but they pack less area per unit volume and cannot be backwashed. Hollow fibres pack far more membrane area into a footprint and can be backwashed to lift internal fouling, but the fibres are more delicate and more sensitive to debris and hair, demanding tighter fine screening upstream.

Whichever geometry is chosen, the selection of the membrane module and its material should follow the feed characteristics — abrasive or fibrous industrial streams favour flat-sheet robustness, whereas a clean municipal feed with tight screening rewards the density of hollow fibre.

How is fouling controlled and CIP scheduled?

Fouling is the accumulation of material that raises the resistance to permeation, and it is the central operating challenge of any MBR. It has three layers, addressed by three defences. The reversible cake of solids on the surface is lifted continuously by air scour and periodic relaxation. The more tenacious layer, removed by daily or weekly in-place chemical dosing, is the pore-blocking and gel fouling caused by extracellular polymeric substances (EPS) and soluble microbial products (SMP). The residual irreversible fouling, which only a full recovery clean restores, is the slow chemical and biological consolidation within the membrane.

Operationally, most plants run at constant flux and watch TMP. As long as TMP creeps slowly the membrane is below its sustainable flux and physical cleaning is keeping pace; a sudden acceleration in TMP means fouling has outrun the air scour and a chemical clean is due. Two chemical regimes handle this. A frequent, mild maintenance clean (sometimes called CEB — chemically enhanced backwash) uses low-concentration sodium hypochlorite for organic fouling and citric or another acid for inorganic scaling, dosed in-place over minutes to hours without removing the membranes. A less frequent, stronger recovery clean or CIP (clean-in-place), at higher chemical concentration and longer soak, restores permeability after months of operation.

Membrane fouling and its cleaning chemistry deserve a full treatment in their own right; our dedicated guide to membrane fouling mechanisms and CIP protocols works through the reagents, concentrations and scheduling in depth — see also membrane fouling and CIP for the full protocol. The design point here is that fouling control is not an afterthought but a set of decisions — scour rate, relaxation cycle, screening fineness, cleaning chemistry — made at the design stage and validated against the sustainable flux, because a membrane run above its sustainable flux will foul irreversibly no matter how aggressive the cleaning.

What footprint and permeate quality does an MBR deliver?

The two outcomes that justify an MBR are a small footprint and an exceptional effluent. The footprint follows directly from the high MLSS: with three to four times the biomass concentration of a conventional plant, the biological reactor is roughly a third of the volume, and because the membrane replaces the secondary clarifier entirely, the largest single tank of a conventional works disappears. A settling clarifier for the same flow might occupy several hundred square metres; the membrane tanks that replace it are a fraction of that area. For a plant hemmed in by an existing site boundary, or an industrial works retrofitting capacity without new land, that compactness is frequently the deciding factor.

The permeate quality is set by the pore size and the complete solids retention. With no route for suspended solids to escape, effluent turbidity sits below 0.2 NTU and suspended solids are effectively undetectable; BOD is typically under 2–5 mg/L because the long SRT drives near-complete oxidation; and the membrane provides a physical barrier that removes a large fraction of bacteria and protozoa, giving a substantial log-reduction without disinfection chemicals. That combination makes MBR permeate directly suitable for non-potable reuse — irrigation, cooling-tower make-up, process water — and an ideal feed for reverse osmosis in industrial water reuse and zero-liquid-discharge schemes, because the near-zero fouling load protects the downstream RO membranes.

The honest conclusion of any MBR design is a trade. The technology buys reuse-grade water and a small footprint at the price of higher energy — the scour and alpha-penalised aeration together push specific energy to 0.6–1.2 kWh/m³ — plus periodic membrane replacement and a demanding fouling-control regime. Where the effluent must be reused or the site is space-constrained, that price is worth paying; where a robust discharge-quality effluent on an open site is all that is required, a conventional or fixed-film process will usually cost less to own and run. Getting the flux, MLSS and aeration numbers right at the design stage, as set out above, is what keeps the trade honest.

How to size a membrane bioreactor: design sequence

1. Fix the biological design. Set SRT (15–30 d) and MLSS (8–12 g/L) from the load, then compute reactor volume from the biomass mass balance and the food-to-microorganism ratio.
2. Determine peak and average flow. Establish average and peak hydraulic flow including recycles; the peak governs membrane area, the average governs energy.
3. Select the design net flux. Choose a net flux (15–25 LMH average) below the sustainable flux, and confirm the peak flux ceiling can be held for the required duration.
4. Calculate membrane area. A = Q / J at the governing (peak) case; round up to whole modules and add one cassette out of service for cleaning.
5. Size the air scour. Set SAD_m (0.2–0.5 Nm3/m2.h) with cyclic aeration; check SAD_p (10–30) and compute blower power.
6. Size alpha-corrected process aeration. Derive AOTR, apply the alpha factor (~0.4–0.5) and field corrections to get the clean-water SOTR, then size blowers accordingly.
7. Define the fouling-control regime. Set relaxation/backwash cycle, screening fineness, and maintenance-clean and CIP chemistry and frequency against the sustainable flux.