Why Pharmaceutical Wastewater Is a Separate Design Problem
Pharmaceutical production-area wastewater is a fundamentally different design condition from the hospital or municipal streams that dominate the published MBR literature, and treating it as a single envelope is the most common specification error on API plant projects. The MDPI 2023 review of high-strength pharmaceutical dosing in MBRs reports pharmaceutical concentrations of 90–31,000 µg/L in production wastewater near Patancheru, India, against 10⁴–10⁵ ng/L in Canadian WWTP influent and low ng/L to low mg/L in the Swiss hospital pilot of Escher et al. (2021) (MDPI 2023, doi:10.3390/membranes13070650; DOI 10.1021/es203495d). Landfill leachate in southern China has been documented with 37 emerging contaminants at 272–1,780 µg/L, confirming that high-strength pharmaceutical streams are a recognised design condition, not an edge case (MDPI 2023).
The Rubí, Spain 21 L pilot tracked pharmaceutical influent across therapeutic classes (analgesics, lipid regulators, antibiotics, psychiatric, antiepileptic, β-blockers, antihistaminics, anti-ulcer, antidiabetic, diuretic) and showed that removal was highly compound-specific rather than uniform (González et al., PMC1805043). The implication for an EPC specification is direct: any MBR design must start with an influent characterisation by therapeutic class, not a single "COD/BOD" envelope borrowed from municipal design guides.
| Stream type | Pharmaceutical concentration | Source |
|---|---|---|
| API production wastewater, Patancheru, India | 90–31,000 µg/L | MDPI 2023 |
| Landfill leachate, southern China | 272–1,780 µg/L (37 emerging contaminants) | MDPI 2023 |
| Canadian WWTP influent | 10⁴–10⁵ ng/L | MDPI 2023 |
| Swiss hospital pilot, sanitary collection | Low ng/L to low mg/L | Escher et al., DOI 10.1021/es203495d |
What an MBR Actually Does Inside a Pharmaceutical Treatment Train
An MBR couples a suspended-growth bioreactor with a membrane separation step, which is what allows the cited pilot to reach MLSS of up to 20 g/L in a compact 21 L reactor volume — settling limitations no longer cap the biomass concentration the way they do in a clarifier (González et al., PMC1805043). Four removal mechanisms act together inside the same tank: biodegradation, sludge adsorption onto the mixed liquor, membrane retention of particulates and biomass, and limited volatilisation. Hena and Znad, summarised in the MDPI 2023 review, report MBR PPCP removals spanning 28–99.8% across these mechanisms.
The high solids retention time is the lever that separates MBR performance from conventional activated sludge for pharmaceuticals. A 30-day SRT retains slower-growing specialist biomass inside the reactor and drives the higher biodegradation that distinguishes MBR from CAS, where shorter SRTs wash out the same organisms. The Rubí pilot reported 87% diclofenac removal in the MBR against about 50% in the parallel CAS train, attributed to the older sludge age (González et al.).
Membrane retention adds a separate removal step for hydrophobic APIs that adsorb onto the cake layer, but hydrophilic APIs rely mainly on biodegradation. This split is the underlying reason that overall pharmaceutical removal is compound-specific rather than uniform, and it is the reason the next sections need to address membrane material and operating envelope separately. PD CEN/TR 15897 treats carbon removal as a baseline requirement and complete nitrification as recommended, which sets the biological floor before any pharmaceutical-specific targets are layered on (BSI, PD CEN/TR 15897).
Submerged MBR Configuration: PVDF Flat-Sheet vs Hollow-Fibre vs Ceramic
PD CEN/TR 15897 covers submerged MBR for municipal wastewater and explicitly states that interchangeability guidelines apply to entire membrane systems rather than individual modules, which is why large submerged MBR systems above 10,000 m³/d tend to use separate membrane tanks rather than integrated cassettes (BSI, PD CEN/TR 15897). The same CEN/TR notes that hollow-fibre and flat-sheet designs can be covered under a single guideline set, since the operational factors that matter — short circuits, EPS build-up, and incomplete biodegradation — are common to both geometries.
For a pharmaceutical application the material question is dominated by fouling behaviour under high-strength feeds. The 2023 Tsinghua ceramic-MBR study used a flat-sheet Al₂O₃ ceramic membrane (0.1 µm pore, 0.043 m² area) operated at 15 L/m²·h with 8/2 min filtration/relaxation, MLSS 6–6.5 g/L, HRT 11 h and SRT 30 d, and documented faster fouling in the reactor dosed with 500 µg/L of five pharmaceuticals (ofloxacin, sulfamethoxazole, sulfamethylthiadiazole, carbamazepine, naproxen) compared with the undosed control (MDPI 2023). EPS accumulation under pharmaceutical stress is the same root cause flagged by PD CEN/TR 15897 as a filterability risk.
For PVDF submerged service, the engineering choice is typically between an integrated package and a module that drops into an existing biological tank. An integrated MBR system with submerged PVDF membranes and a documented footprint reduction of 60% against conventional layouts covers 10–2,000 m³/d. For retrofit or larger flows, a DF-series PVDF flat-sheet module at 0.1 µm, 80–225 m² per module and 32–135 m³/d per module is the geometry that aligns with the CEN/TR separate-tank convention once flows cross the 10,000 m³/d threshold.
| Option | Membrane / area | Flux / cycle | MLSS, HRT, SRT | Best fit |
|---|---|---|---|---|
| PVDF integrated package | Submerged PVDF, <1 µm effluent | Per project | Project-specific | 10–2,000 m³/d, greenfield |
| PVDF flat-sheet module (DF series) | 0.1 µm, 80–225 m² | 32–135 m³/d per module | Project-specific | Retrofit membrane tank, ≥10,000 m³/d trains |
| Ceramic flat-sheet (Al₂O₃, MDPI 2023) | 0.1 µm, 0.043 m² | 15 L/m²·h, 8/2 min | MLSS 6–6.5 g/L, HRT 11 h, SRT 30 d | High-strength feed, 180-day stable operation |
Design Parameters That Move Pharmaceutical Removal
The 2023 ceramic-MBR dataset is the most defensible 180-day operating envelope currently available for MBRs under pharmaceutical stress, and it is the right baseline for an EPC design. With SRT 30 d, HRT 11 h, MLSS 6–6.5 g/L and a 300% recirculation ratio across the anoxic–aerobic–membrane layout, the dosed reactor (MBRe) achieved >98% DOC removal and 90.1% TN removal over 180 days (MDPI 2023). Phosphorus removal fell from 72.6% in the undosed control to 57.8% under pharmaceutical stress, which is the most sensitive biological link in the train and the parameter most likely to slip on a real API plant feed.
Membrane flux was held at 15 L/m²·h with 8/2 min filtration/relaxation and chemical cleaning triggered at ΔTMP = 30 kPa, which was sufficient to maintain operation constant over 180 days even with accelerated fouling in the dosed reactor (MDPI 2023). The Rubí pilot used the same intermittent permeation logic — 8 min on, 2 min halt — to limit cake compaction in pharmaceutical service (González et al.). PD CEN/TR 15897 treats carbon removal as a baseline requirement and complete nitrification as recommended, which sets the floor for biological design before any pharmaceutical-specific targets are layered on (BSI).
| Parameter | Operating point | Evidence (source, year) |
|---|---|---|
| SRT | 30 d | MDPI 2023 (Li et al.) |
| HRT | 11 h | MDPI 2023 (Li et al.) |
| MLSS | 6–6.5 g/L (ceramic); up to ~20 g/L reported for MBR generally | MDPI 2023; González et al. |
| Flux | 15 L/m²·h, 8/2 min filtration/relaxation | MDPI 2023 (Li et al.) |
| Cleaning trigger | ΔTMP = 30 kPa | MDPI 2023 (Li et al.) |
| Recirculation | 300% across anoxic–aerobic–membrane | MDPI 2023 (Li et al.) |
| DOC removal | >98% (control and dosed) | MDPI 2023 (Li et al.) |
| TN removal | 90.1% (dosed) vs 83.2% (control) | MDPI 2023 (Li et al.) |
| TP removal | 57.8% (dosed) vs 72.6% (control) | MDPI 2023 (Li et al.) |
The Persistent-Compound Problem: Where MBR Alone Is Not Enough
Carbamazepine is the canonical persistent API for MBR design, and the numbers are not close. In the Rubí pilot, carbamazepine removal was <20%, with effluent concentrations frequently above the influent because of conjugation-deconjugation during the biological step (González et al., PMC1805043). The MDPI 2023 review confirms that carbamazepine removal in MBRs ranges from negative values up to 23% and lists sulfamethylthiadiazole as the most recalcitrant sulfonamide with negligible removal, while sulfamethoxazole spans 20–92% depending on operating conditions (MDPI 2023).
Even within a single therapeutic class the variation is wide enough to break any blanket "MBR removes X%" claim. Ofloxacin removal in MBRs is reported at 33.9–95.2% and naproxen across a comparable range, which is operating-condition driven, not chemistry driven (MDPI 2023). The Swiss hospital pilot tracked 68 target analytes including 56 pharmaceuticals across antibiotics, antimycotics, antivirals, iodinated X-ray contrast media, anti-inflammatories and cytostatics, and MBR alone did not deliver all compounds below detection (Escher et al., DOI 10.1021/es203495d).
For an API plant the practical conclusion is to specify MBR as a high-efficiency biological pre-treatment with a polishing step sized to the worst few persistent compounds, not as the final compliance barrier. Carbamazepine, sulfamethylthiadiazole, sulfamethoxazole at the high end, and iodinated X-ray contrast media are the compounds that have to be addressed downstream with AOP, ozone, granular activated carbon, or RO. The MDPI 2023 study also flagged faster fouling under pharmaceutical stress, so the polishing step should be designed to take an MBR permeate that may already carry higher SMP and EPS loading than a municipal feed.
Decision Framework: When an MBR Earns Its Place in a Pharma Train
Convert the data above into four procurement-grade rules of thumb before sizing equipment:
- Choose MBR as the biological core when the plant needs high MLSS (the Rubí MBR reached 20 g/L), a small footprint, and a reuse-quality supernatant suitable for RO polishing.
- Choose separate membrane tank over integrated when design flow approaches or exceeds 10,000 m³/d, following the PD CEN/TR 15897 convention for large submerged MBRs (BSI).
- Specify a polishing step whenever the influent characterisation includes persistent APIs such as carbamazepine, sulfamethylthiadiazole, or iodinated X-ray contrast media; MBR alone will not meet tightening effluent limits on these compounds.
- Specify an anoxic–aerobic–membrane layout with SRT ≥ 30 d when simultaneous nitrification and pharmaceutical biodegradation are required, based on the 2023 ceramic-MBR evidence base (MDPI 2023).
For reuse or tight discharge envelopes, the polishing train typically pairs an industrial RO system downstream of the MBR, sized against the MBR permeate quality rather than the raw influent. For sizing the polishing block itself, the AOP system design guide for 2026 and the 2026 CAPEX and OPEX cost benchmarks per MGD provide the next layer of detail. Where the driver is plant-wide water recovery rather than discharge compliance, the manufacturing water-reuse engineering guide gives the system-level context.
Frequently Asked Questions
What CAPEX and OPEX should we budget for an MBR on a pharmaceutical wastewater train?
Specific MBR capital and operating cost figures depend on influent characterisation, flow, the MBR permeate quality target, and the chosen polishing train (AOP, ozone, GAC or RO). For a defensible budget, request from each bidder a side-by-side statement of design flow, peak flow, SRT, MLSS, membrane area, specific air demand per membrane, chemical-cleaning regime, and energy use per m³ of permeate, then validate against the 2026 CAPEX and OPEX cost benchmarks per MGD.
How do I choose between suppliers of submerged MBR systems for an API plant?
Score bidders against three things the research actually supports: (1) evidence of stable operation at SRT ≥ 30 d under pharmaceutical stress rather than only municipal data, (2) PD CEN/TR 15897-compliant documentation that addresses EPS build-up, short circuits and incomplete biodegradation, and (3) a polishing reference list (AOP/ozone/RO) sized for persistent APIs such as carbamazepine, sulfamethylthiadiazole and iodinated X-ray contrast media. Treat the absence of pharmaceutical-specific operating data as a compliance risk for the bid review.
Is ceramic or PVDF the better membrane material for pharmaceutical MBR service?
The 2023 Tsinghua study used a 0.1 µm Al₂O₃ flat-sheet ceramic membrane and maintained stable operation for 180 days at 15 L/m²·h with 8/2 min filtration/relaxation under 500 µg/L pharmaceutical dosing, while documenting faster fouling in the dosed reactor (MDPI 2023). PVDF submerged designs in the 0.1 µm flat-sheet class are the more common industrial configuration and are covered by the same PD CEN/TR 15897 framework, so material choice is driven by feed chemistry, CIP temperature limits and lifecycle cost rather than by inherent performance for the API compounds covered in the public data.
Do we still need AOP or RO polishing if we run an MBR at SRT 30 d?
Yes, for the persistent compound set. Carbamazepine removal in the Rubí MBR was <20% and the MDPI 2023 review puts MBR carbamazepine removal at negative to 23%, with sulfamethylthiadiazole reported as negligible and ofloxacin at 33.9–95.2% (González et al.; MDPI 2023). An MBR at SRT 30 d reliably delivers >98% DOC removal and ~90% TN removal, but polishing has to be specified and sized against the worst few APIs in the influent characterisation, not against DOC.
