Why Cooling Tower Blowdown Is the Hardest Stream in the Plant
Blowdown is the controlled bleed that keeps dissolved solids, silica, and treatment-chemistries from concentrating past their solubility limits as makeup water evaporates from the cooling loop. Each kilogram of evaporation pulls nothing out, so cycles of concentration (CoC) are the operating dial: at 4 CoC, a 100 mg/L Ca feed becomes 400 mg/L in the loop; at 6 CoC, 600 mg/L — and that curve is what decides how much water you bleed. Cooling water accounts for 50–55% of all water used in a refinery (PMC9029438), so blowdown is typically the largest single reuse or discharge lever on an industrial site.
The 1973 OSTI K-4023 softening study remains the cleanest field data point on the problem: tower blowdown at 450 ppm Ca as CaCO₃ was routinely softened to 50 ppm Ca at room temperature, and the historical operating ceiling for the recirculating water was 300 ppm Ca as CaCO₃ (OSTI K-P-4023, 1973). Those numbers frame the rest of the train. A modern blowdown typically presents 1,500–5,000 mg/L TDS, 50–200 mg/L silica, residual oxidizing biocides (Cl₂, ClO₂, bromine), phosphate and sulfate scale formers, and temperatures of 32–45°C. None of these is exotic on its own, but in combination they produce a stream that fouls membranes, scales RO, and is hot enough to destabilize biological floc.
The design consequence is non-negotiable: a biological MBR does not desalt. It strips organics, oxidizer-damaged biomass, and silica-bound colloids, but TDS, chloride, and the bulk hardness leave the bioreactor untouched. Any reuse train that returns water to the cooling loop — or hits a strict chloride/TDS discharge limit — must pair MBR with softening, RO, or membrane capacitive deionization (MCDI) downstream.
Submerged vs Sidestream MBR: Which Configuration Fits Blowdown
Submerged MBR with PVDF hollow-fiber or flat-sheet modules is the 2026 default for cooling tower blowdown: the membranes sit in the aeration tank, scouring air keeps fouling under control, and energy use runs an order of magnitude below cross-flow designs. The boundary conditions for this configuration are TSS <200 mg/L, TDS <3,000 mg/L, and silica <150 mg/L. Above those numbers, aeration scours lose to fouling rate and a sidestream configuration earns its energy penalty back in stable flux.
Sidestream (external cross-flow) MBR pushes the mixed liquor through a separate loop at high cross-flow velocity, typically 1–3 m/s. It tolerates higher TSS, higher temperature (up to ~45°C steady), and higher TDS, at 0.5–1.5 kWh/m³ of incremental energy over submerged operation. The trade is real estate: the loop, the recirculation pump, and the higher shear header add footprint and mechanical complexity. For most data-center and HVAC loads, where blowdown runs cool and low-strength, submerged wins on capex and opex. For refinery cooling loops with hot, silica-laden, high-TDS blowdown, sidestream is the safer call.
Inside the submerged envelope, the choice is hollow fiber versus flat sheet. Hollow fiber packs more membrane area per cassette and runs cheaper per m², but is more vulnerable to fouling from hair, fiber, and scale-chip carryover because the bundle cannot be opened up. Flat sheet — including the DF series flat-sheet membrane modules at 0.1 μm pore — tolerates aggressive backwash, is field-replaceable element by element, and gives the operator a visible, inspectable surface. Cooling blowdown at 0.1 μm does not need tighter UF (0.04 μm) unless an RO vendor specifies it for Silt Density Index (SDI) margin; MBR effluent at SDI <3 is the typical target.
The reference benchmark for this configuration as an RO pretreatment is Tam LS et al. and the broader MBR/RO pilot literature summarized in J Environ Health Sci Eng 11:34, which documents MBR/RO and MF/RO reclamation pilots with consistent >90% COD removal and stable downstream RO operation.
| Parameter | Submerged MBR (PVDF HF/FS) | Sidestream Cross-Flow MBR |
|---|---|---|
| Best-fit TDS | <3,000 mg/L | 3,000–8,000 mg/L |
| Silica tolerance | <150 mg/L with pretreatment | <250 mg/L |
| Max operating temperature | ~40°C | ~45°C steady |
| Energy use | 0.05–0.3 kWh/m³ | 0.5–1.5 kWh/m³ |
| Footprint | Low (membranes in aeration tank) | Higher (external loop + recirculation pump) |
| Maintenance | In-tank inspection, cassette lift | Loop disassembly, pump service |
| Typical pore size | 0.1 μm (flat sheet) / 0.04 μm (HF UF) | 0.1–0.2 μm tubular |
| Field-replaceable elements | Yes (flat sheet) | Element-by-element on most tubular designs |
Pretreatment Train Before the MBR
An MBR protects itself from biomass, not from chemistry. The pretreatment train is what keeps silica, hardness, oxidizer residual, and debris from collapsing flux, so this section deserves the same engineering effort as the membrane selection itself.
Step 1 is screening. A 2–5 mm rotary bar screen headwork stops tower debris, scale chips, and macro-fouling material that would otherwise pin to fibers or wedge into module headers. Step 2 is partial lime-soda softening to the OSTI baseline: Ca from ~450 ppm to 50 ppm, and silica dropped to <100 mg/L to protect both MBR flux and the downstream RO (OSTI K-P-4023, 1973). Step 3 is chemical conditioning via a PLC-controlled chemical dosing skid: anti-scalant, sodium bisulfite (NaHSO₃) or activated carbon for free chlorine reduction (target Cl₂ residual <0.1 mg/L at the membrane), and pH adjustment to 6.5–7.5 to keep PVDF within its stable operating window. Step 4 is cooling and equalization: drop the stream to <35°C so biological kinetics and viscosity stay predictable, and dampen the flow spikes that intermittent blowdown produces.
The NREL/Aqualogix BDR project (NREL/TP-2C00-86615, Aug 2023) validated exactly this approach — a membrane array was retrofitted onto existing partial softening rather than replacing the legacy chemical program, and water/sewer savings were the metered performance objective. The takeaway for 2026 designs: do not throw out a working softener to install a membrane. Wrap the membrane around the softener, dose conditioning, and let the biological step handle what softening does not.
| Stage | Equipment | Target / Setpoint | Function |
|---|---|---|---|
| 1. Screening | Rotary bar screen (2–5 mm) | <5 mm opening | Remove debris, scale chips, macro-foulants |
| 2. Softening | Lime-soda or partial ion exchange | Ca: 450 → 50 ppm; SiO₂ <100 mg/L | Hardness and silica control |
| 3. Chemical conditioning | PLC dosing skid (anti-scalant, NaHSO₃, pH) | Cl₂ <0.1 mg/L; pH 6.5–7.5 | Protect PVDF, condition RO feed |
| 4. Cooling & EQ | Plate heat exchanger + equalization basin | T <35°C; 6–12 hr HRT buffer | Stable flux, dampen flow spikes |
MBR Effluent Specs and How They Enable Reuse or Discharge
An MBR delivers a specific, defensible effluent envelope: COD removal >90%, BOD₅ <5 mg/L, TSS <5 mg/L, and turbidity <1 NTU (per the MBR/RO pilot literature in J Environ Health Sci Eng 11:34). Those numbers satisfy the biological parameters of EPA secondary discharge limits, the EU Urban Waste Water Directive 91/271/EEC, and most provincial GB tables — but they do not move the needle on TDS, total nitrogen, or chloride. If the discharge permit is bracketed by salinity, the MBR alone does not close the compliance gap.
For reuse, MBR effluent at SDI <3 is exactly what an RO wants to see. Industrial RO trains routinely run 70–85% recovery on MBR-polished feed, and the concentrate — typically 5–15% of the flow at 5,000–15,000 mg/L TDS — gets routed to an MCDI cell, a brine concentrator, or a ZLD crystallizer. MCDI single-pass performance on cooling tower blowdown has been demonstrated in high-tech industry settings (ScienceDirect S0921344921006200), making it a credible bridge between RO and full ZLD.
The economic inflection point is the freshwater offset. When potable or clarified-river makeup exceeds roughly $2/m³ and sewer surcharges stack on top, an MBR + RO polishing train pays back the membrane capex in 3–5 years. Below that, the MBR alone is the rational play: minimum capex, biological discharge compliance, no RO membrane to clean. The industrial RO polishing step is the right call only when the reuse target is the cooling loop itself or when discharge limits are chloride-driven.
| Parameter | MBR Influent (Cooling Blowdown) | MBR Effluent | RO Polished (post-MBR) |
|---|---|---|---|
| COD | 100–500 mg/L | <50 mg/L (>90% removal) | <10 mg/L |
| BOD₅ | 20–80 mg/L | <5 mg/L | <2 mg/L |
| TSS | 50–200 mg/L | <5 mg/L | <1 mg/L |
| Turbidity | 10–50 NTU | <1 NTU | <0.1 NTU |
| TDS | 1,500–5,000 mg/L | Unchanged (no desalination) | <500 mg/L (75% recovery baseline) |
| SDI | >6 | <3 | <1 |
| Free Cl₂ | 0.5–2 mg/L | <0.1 mg/L (post-reduction) | — |
2026 Selection Framework: MBR, MBBR, or Direct RO
Three influent parameters decide the configuration: biological load, TDS, and silica. Anything else is detail.
If biological load is low (COD <60 mg/L) and TDS is moderate (<1,500 mg/L), MBBR is cheaper than MBR; skip to RO polishing. If biological load is moderate-to-high (COD 60–500 mg/L, ammonia, organics from biocide degradation), submerged MBR is the 2026 default — biological polishing and a membrane barrier in one tank. If TDS >3,000 mg/L, silica >150 mg/L, or temperature >40°C, choose sidestream MBR or pre-soften aggressively before any submerged system.
For data centers in freshwater-stressed grids, the calculus is moving. In Northern Virginia, where 2025–2026 utility rates and sewer surcharges have made reuse financially unavoidable, the MBR + RO train is becoming the standard 2026 reuse package, with ZLD on the RO concentrate to eliminate liquid discharge entirely. Dublin is following the same curve, with monitoring design tightening on ammonia and chloride as the Uisce Éireann discharge framework evolves. The MBBR alternative for data center cooling blowdown still has a place at low-COD sites, but new hyperscale builds are landing on submerged MBR as the baseline. For a complete submerged PVDF MBR system sized to this duty, the engineering flow is softening → equalization → MBR → RO → concentrate management.
Frequently Asked Questions
What influent hardness should the pretreatment target before the MBR?
Drop Ca from 450 ppm to 50 ppm as CaCO₃ and hold silica below 100 mg/L (OSTI K-P-4023, 1973 baseline).
What COD removal does an MBR actually deliver on cooling blowdown?
Greater than 90% COD removal, with BOD₅ under 5 mg/L and TSS under 5 mg/L (per MBR/RO pilot data, J Environ Health Sci Eng 11:34).
Submerged or sidestream MBR for high-silica refinery blowdown?
Sidestream cross-flow MBR when silica exceeds 150 mg/L or TDS exceeds 3,000 mg/L; otherwise submerged PVDF flat-sheet.
When does RO polishing become economically justified after the MBR?
When freshwater/sewer offsets exceed roughly $2/m³, which is the typical 2026 threshold for industrial reuse payback.
What is the most common MBR fault on cooling blowdown duty?
Oxidizer shock from residual Cl₂ or ClO₂ — the seven data-backed fixes are in the MBR troubleshooting playbook.
How is Dublin monitoring design changing for cooling blowdown reuse in 2026?
Continuous ammonia and chloride online monitoring is becoming standard; see the Dublin cooling blowdown monitoring design guide.
What pairs with RO concentrate from a blowdown reuse train?
MCDI, brine concentrator, or full ZLD crystallizer; the ZLD concentrate management for cooling blowdown guide covers the engineering.
