MBR Monitoring: MLSS, Fouling and the Sensors That Save Membranes

High-traffic-inspired · Technical · Published: October 14, 2026

Key Takeaways

  • In a membrane bioreactor, the membrane is only as healthy as the mixed liquor around it. MBRs run MLSS concentrations around 8–12 g/L, and sludge viscosity rises sharply across exactly that band [1].
  • Fouling under real operating conditions progresses through three TMP stages: a slow linear rise (stage I), an exponential climb (stage II), then a rapid linear jump (stage III) once the cake layer covers the membrane [2].
  • The monitoring job is to act inside stage II. Once TMP enters stage III, effective filtration area is collapsing and only intensive chemical cleaning or recovery cleaning brings it back [2][3].
  • Design guidance for submerged systems puts net membrane flux at 10–25 LMH with peak flux capped near 30 LMH, MLSS at 8,000–12,000 mg/L, and sludge retention times of 20–40 days; above roughly 15,000 mg/L, oxygen transfer efficiency drops sharply [3].
  • The sensor set that saves membranes is small: mixed-liquor MLSS and viscosity trends, TMP slope, permeate turbidity (alarm near 0.2 NTU, auto-divert above it), and biological-tank DO held between its commissioning alarms — 1 mg/L low and 7 mg/L high in the worked example [4].
  • Plants that keep TMP inside a disciplined band (roughly 10–30 kPa on many flat-sheet installations) and clean on schedule routinely report membrane service lives of five to seven years. TMP left unwatched shortens that to a fraction [5].

Why MBR Monitoring Is Different

A membrane bioreactor merges two processes that conventional plants keep apart: biological treatment and membrane solid-liquid separation. The bioreactor and membrane stages cannot be treated as separate unit operations, because they interact through sludge properties — the same mixed liquor that feeds the biology also feeds the fouling. That interaction is why MBR monitoring is a sensors story before it is a chemistry story. The operator’s daily job is to hold a narrow hydraulic and biological window, and instruments are the only way to see it in time.

The stakes favor prevention, by a wide margin. A membrane cassette replacement is a capital event; a drum of cleaning chemical is not. Every monitoring decision should be judged by one question: did it let us act during stage II?

The Mixed Liquor: MLSS, Viscosity and the 10 g/L Question

MBRs run concentrated on purpose. Design guidance for submerged systems holds biomass at 8,000–12,000 mg/L MLSS to shrink the reactor and extend sludge age, at the cost of thicker, more viscous liquor [3]. The viscosity penalty is not linear. Laboratory and full-scale filterability studies report that sludge viscosity increases dramatically across the 8–12 g/L band and worsens filtration, and that aeration intensity only stops compensating once concentrations climb past about 10 g/L [1]. One synthesis of MBR fouling studies places the critical point at MLSS of 10 g/L: above it, sludge viscosity rises across a wide range and membrane filterability suffers measurably [1].

This produces a paradox that trips up new plants: the highest MLSS is not the best MLSS. Biomass concentration is a controlled variable, not a scoreboard. Keep it inside the design window, waste sludge consistently to hold the SRT target, and watch viscosity-trending parameters rather than pushing concentration for its own sake [3].

Monitoring parameter Target / threshold Action when crossed
MLSS in membrane tank 8,000–12,000 mg/L (design window) Adjust sludge wasting; review loading; verify SRT [3]
Sludge viscosity / filterability trend Stable, no sustained rise Increase wasting; check air scour; review fouling drivers [1]
TMP at constant flux Slow linear rise only (stage I–II) Trigger relaxation/backpulse review; plan maintenance cleaning [2]
TMP slope acceleration Exponential or rapid-linear rise (stage II–III) Maintenance clean now; prepare recovery clean before stage III [2]
Permeate turbidity High alarm near 0.17 NTU; auto-divert at 0.2 NTU Bubble test and repair the suspect cassette [4]
Biological tank DO 1 mg/L low alarm / 7 mg/L high alarm (worked commissioning example) Trim blower output or fix oxygen transfer (viscosity, fouled diffusers) [4]
Net flux 10–25 LMH design; ≤30 LMH peak for submerged systems Throttle permeate pump during peak-flow events [3]

TMP and the Three Stages of Fouling

Transmembrane pressure is the most honest fouling indicator because it integrates everything: pore blocking, gel layer, cake, temperature and flux. Under sub-critical flux operation, fouling studies show the TMP trajectory decomposes into three stages — a slow linear increase (stage I), an exponential increase (stage II), and a final rapid linear rise (stage III) when the entire membrane surface is covered by a cake layer [2].

Read it operationally: stage I is background biology, stage II is the warning window, stage III is the invoice. A plant that logs TMP at constant flux and alarms on slope — not on absolute value alone — buys itself the stage-II intervention window. Maintenance cleaning scheduled inside that window keeps the membrane from ever seeing stage III. Plants that hold TMP inside a disciplined band and run chemically enhanced backwashes on schedule report flat-sheet and hollow-fiber service lives in the five-to-seven-year range [5].

Flux, Relaxation and Backwash Strategy

Because MBRs run at constant flow, flux discipline is a control problem. Submerged-system design guidance puts net flux at 10–25 LMH and recommends designing at 80% of tested peak, with peak-flow events capped near 30 LMH [3]. Above the critical flux, TMP climbs quickly and part of the fouling becomes irreversible — the operating manual’s job is to keep the system below it [3].

Between chemical cleans, the hydraulic tools are simple and underused:

  • Relaxation. Pause permeate extraction for seconds to minutes (30 s to 2 min is typical practice) while air scour keeps lifting cake off the fibers [3]. Common duty cycles pause the permeate pump briefly at intervals so the scouring air can do its work.
  • Backpulse / backwash. Short reverse-flow pulses dislodge particles embedded at the membrane face; frequency should match the observed fouling rate, not the factory default [3].
  • Seasonal compensation. Water viscosity climbs as temperature falls, so lower the flux in cold seasons to match the colder, thicker mixed liquor.

None of these strategies can be scheduled blind. They need the TMP trend, the MLSS trend and temperature in one synchronized data stream — which is exactly what an integrated monitoring loop provides.

The Sensor Set That Saves Membranes

Four instrument families carry the protection burden:

  1. MLSS / suspended-solids sensor in the membrane tank. Guards the viscosity window and drives wasting decisions; the same measurement validates that air-scour energy matches the sludge actually present [1][3].
  2. TMP pair (feed-side and permeate-side pressure) with slope alarming. The three-stage model gives operators a mathematical early warning: alarm on d(TMP)/dt at constant flux, not on a fixed threshold reached too late [2].
  3. Permeate turbidity meter. The membrane’s own integrity gauge. Online turbidity on the permeate line detects breaches instantly, with commissioning guidance setting a high alarm near 0.17 NTU and an auto-divert at 0.2 NTU, sending effluent to the sewer while the suspect skid is bubble-tested and repaired [4].
  4. DO (and pH) in the biological tanks. Aeration is the largest energy consumer and the fouling-strategy enabler. Holding DO between the low and high alarms defined at commissioning — 1 and 7 mg/L in the worked example — protects nitrification without over-aerating a viscous, concentrated liquor [4].

Integration matters more than any single probe. Multi-parameter probes that combine pH, ORP, conductivity and temperature in one body cut the port count and calibration load — four parameters, one probe, zero alignment error — and digital Modbus RTU/TCP outputs push synchronized streams to SCADA or predictive models, sensors that feed your AI water model, not just your dashboard [6]. When the MLSS trend, the TMP slope and the DO curve share one time base, stage II announces itself weeks before a cleaning emergency.

A Week-One Checklist for New MBR Operators

  1. Post the design window on the wall: MLSS band, net flux cap, TMP alarm slopes, DO alarm band [3][4].
  2. Verify every alarm does what the commissioning document says — especially the permeate-turbidity auto-divert [4].
  3. Log TMP at the same time daily under comparable flux; slope beats absolute value [2].
  4. Watch viscosity proxies (MLSS, filterability, air-scour distribution) rather than chasing a single number [1].
  5. Clean on schedule, not on crisis; record TMP recovery after each clean as the membrane’s health ledger [5].

Membranes rarely fail suddenly. They fail by inches, in stage II, while the dashboard shows an average. The sensor set above is cheap insurance against that invoice.

References

  1. TU Delft OpenCourseWare — “Filterability and Sludge Concentration in Membrane Bioreactors.” Reviews MLSS–fouling literature: dramatic viscosity increase across the 8–12 g/L band, sharply reduced effectiveness of aeration compensation above ~10 g/L, and the 10 g/L critical point cited in fouling syntheses. https://ocw.tudelft.nl/wp-content/uploads/Filterability_and_Sludge_Concentration_in_Membrane_Bioreactors.pdf
  2. Li, J., Zhang, X., Cheng, F., Liu, Y. (2013) — “New insights into membrane fouling in submerged MBR under sub-critical flux condition,” Bioresource Technology (PubMed 23611700). Establishes the three-stage TMP progression: slow linear (stage I), exponential (stage II), rapid linear (stage III) with full cake coverage. https://pubmed.ncbi.nlm.nih.gov/23611700/
  3. Spans Envirotech — “Membrane Bioreactor (MBR) for Industrial ETPs — CPCB Design Guidelines.” Design envelope: net flux 10–25 LMH submerged, ≤30 LMH peak, MLSS 8,000–12,000 mg/L with sharp oxygen-transfer loss above ~15,000 mg/L, SRT 20–40 days, and a three-tier cleaning protocol including 30 s–2 min relaxation. https://spans.co.in/knowledge/cpcb-etp-design-guidelines-mbr/
  4. San Francisco Public Utilities Commission — “Guidebook for Commissioning an Onsite Water Treatment System” (2022). Example MBR control points: instantaneous filtrate turbidity high alarm 0.17 NTU / hi-hi 0.2 NTU with automatic diversion to the sewer; bioreactor DO low alarm 1 mg/L, high 7 mg/L; pH 6–8. https://www.sfpuc.gov/sites/default/files/documents/CommissioningOnsiteWaterReuse_2022_Final.pdf
  5. HydropureWater — “Flat Sheet MBR Membrane Maintenance Protocols 2026.” Reports service life of 5–7 years when TMP stays inside 10–30 kPa, chemically enhanced backwash three times per week, and offline soak above 35–40 kPa. https://hydropurewater.com/blog/670-flat-sheet-mbr-membrane-maintenance-guide-protocols-for-peak-performance-longevity.html
  6. Shanghai ChiMay — 8 Essential Parameters Every Plant Must Monitor for Water Quality (technical resource). Sensor technology, typical accuracy and maintenance intervals for conductivity, pH, dissolved oxygen, turbidity and related parameters, with integrated transmitter systems for plant networks. https://shchimay.com/en/8-essential-parameters-every-power-plant-must-monitor-for-water-quality/

About the author: Written by the Shanghai ChiMay Technical Editorial Team — process engineers who design and support in-line water quality analyzers and multi-parameter sensors for membrane-based wastewater and water-reuse systems.