Why Kolkata data centers face a different blowdown problem in 2026
A 100 MW hyperscale data center draws up to 2 million liters of water per day, roughly the daily use of thousands of households (per IDE-Tech, 2026), and in Kolkata that volume is pulled from a stressed basin with a regulator that is no longer patient. The Hooghly River and the KMWC municipal distribution feeding Salt Lake Sector V and Rajarhat swing from ~120 mg/L TDS in the post-monsoon window to >350 mg/L through March–June (Hooghly basin monitoring, 2025-08), so the cycles of concentration a cooling tower can safely hold — and the chemistry of the blowdown it produces — change with the season. The same metros that are absorbing the 2026 hyperscale build pipeline (SALT LAKE Sector V, Rajarhat New Town) draw from the same sewer shed, which is why the West Bengal Pollution Control Board is tightening Consent to Operate conditions on new IT parks in the basin. CPCB Schedule-I general standards for discharge into surface water and Schedule-VI effluent tolerances are treated here as binding constraints, not optional targets (per CPCB, 2025-11), and any Ganga-basin facility can be pushed toward a near-zero-discharge condition by WBPCB if freshwater intake is capped or reuse targets are unmet. The practical consequence: a generic global CTBD playbook misses two things that matter in Kolkata — a feed water that swings seasonally and a regulator that will rewrite the consent if the design does not already maximize reuse.
What a Kolkata data center actually has to treat: feed streams and their chemistry
Before picking equipment, separate the campus into the streams you actually have to handle. Conflating them is the single most common reason an Indian data center wastewater treatment plant gets oversized upstream and undersized at the membranes.
| Stream | Typical source | TDS (mg/L) | COD (mg/L) | BOD (mg/L) | TSS (mg/L) | Key design driver |
|---|---|---|---|---|---|---|
| Domestic sewage (staff blocks, kitchens) | Restrooms, pantries, rest areas | 500–1,200 | 250–500 | 150–300 | 200–400 | MBR sizing, peak shift loading |
| Cooling tower blowdown (CTBD) | Evaporative cooling loop | 1,200–6,000 | 20–80 | <10 | 10–50 | Silica, CaCO3, CaSO4 scaling; biocides |
| RO reject (makeup plant) | BWRO on Hooghly/KMWC | 2,000–5,000 | 10–30 | <5 | 5–15 | Brackish baseline, scales with intake TDS |
| DG cooling / humidification bleed | Diesel plant, humidifiers | 500–2,000 | 30–100 | 10–30 | 10–30 | Oil/grease from DG; intermittent flow |
| Floor wash & equipment drain | UPS rooms, plant rooms | 200–800 | 50–200 | 20–80 | 20–100 | Spill risk, hot spots |
CTBD is the dominant stream by volume and the most chemically constrained. It typically carries 1,200–6,000 mg/L TDS — about 4–8× the makeup water concentration — with the scaling fraction dominated by silica, calcium carbonate, and calcium sulfate, plus accumulated biocides, corrosion inhibitors, and dispersants from the cooling water program (per Genesis Water Tech, 2026). RO reject from the makeup water plant is brackish and tracks Hooghly intake TDS. Domestic sewage, humidification bleed, and DG cooling are smaller and intermittent but must be characterized before being blended into the CTBD equalization tank, because oil from the DG area or a surfactant slug from kitchen drain will choke a UF rack within hours if it reaches the membranes un-stripped.
The three disposal routes and the 2026 West Bengal / CPCB compliance envelope
Choosing the disposal route is the decision that sets the entire treatment train. Spend the capital conversation here, not on the equipment list.
Discharge to municipal sewer requires a WBPCB Consent to Operate under the Water Act, 1974, with site-specific limits on TDS, BOD, COD, suspended solids, and heavy metals. In practice, WBPCB conditions on hyperscale campuses in the Hooghly basin are already stricter than CPCB Schedule-I defaults (per CPCB Schedule-I, 2025-11) — typical TDS caps sit in the 2,000–2,100 mg/L range and BOD at 30 mg/L or below for sewer discharge. Reuse for cooling-tower makeup falls under CPCB reuse guidelines and the facility's own water-stewardship policy; if the blowdown is fully recycled, no separate SPCB discharge consent is triggered for that stream, though the makeup water plant still requires its own consent. Zero liquid discharge is imposed where the facility sits in a notified Ganga-basin no-discharge zone or where freshwater intake is capped by consent conditions, and it is no longer hypothetical — Kolkata's hyperscale proposals in 2026 are increasingly being cleared on reuse-plus-ZLD rather than on sewer discharge. The LCA caveat matters: reclaimed-water trains carry roughly 2× the GWP of freshwater in today's West Bengal grid, with treatment energy making up ~80% of the gap, but under a 2030+ decarbonized grid that penalty falls below 5% of the freshwater baseline (Cartagena Vaca et al., 2026). In other words, the carbon argument against reuse is a 2025 argument; it will not survive the lifetime of an asset commissioned in 2026.
Pretreatment that protects every downstream membrane
Every rupee spent on pretreatment saves five on membrane replacement and unscheduled CIP. Skipping it is the most expensive line item a Kolkata facility can write into its 2026 design.
Start with rotary bar screening on the cooling basin side-stream, then a multi-media filter sized to take the circulating water from 10–50 mg/L suspended solids down to 10–15 mg/L at the RO feed. Genesis Water Tech (2026) recommends side-stream filtration at 1–5% of circulation flow using 10–25 µm self-cleaning screens — this is the cheapest reliability insurance on the plant. Oily streams from the DG area should pass through a DAF unit before joining the equalization tank, otherwise a fuel or lube-oil upset will pin the UF membranes within an hour. From there, route the stream through an UF pretreatment ahead of RO at 0.01–0.1 µm pore size, operating at 10–30 psi with permeate backwash and CIP every 1–3 months at 90–95% recovery. Antiscalant selection must be matched to the blended CTBD + RO reject chemistry, not the makeup water alone — dose to a Langelier Saturation Index target and verify silica scaling potential against the projected top-of-cycle concentrate, especially in the March–June window when Hooghly TDS and silica both peak. HydropureWater's UF CAPEX and OPEX guide is a useful sizing reference for this block.
High-recovery RO and the case for controlled salt precipitation
Conventional brackish water RO plateaus at 75–80% recovery on CTBD before silica, CaCO3, and CaSO4 force a shutdown (per IDE-Tech, 2026). That ceiling is fine for generic reuse, but it leaves a concentrate stream that is too large and too scaling-prone to send to a downstream MVC or crystallizer economically. Pushing past it requires a different philosophy.
The first pass runs a high-recovery industrial RO system at a conservative 50–70% local recovery. Permeate comes off at 10–50 mg/L TDS, well inside cooling-tower makeup spec (per Genesis Water Tech, 2026). The concentrate is not sent to a second RO stage or to disposal; it is routed to a fluidized-bed crystallizer where the antiscalant is intentionally deactivated and silica, CaCO3, and other sparingly soluble salts precipitate onto seed material as compact pellets. Once the scaling species are pulled out of solution, the remaining brine is a near-pure NaCl stream that a single additional RO stage can process at very high recovery — the IDE-Tech case study (2026) reports ~95% overall recovery and ~1 mg/L silica in the permeate. The RO stages themselves are run in a dynamic mode: short production periods alternating with high-velocity flushes that keep the membrane surface inside the induction phase of crystallization, extending CIP intervals and removing the need for interstage boosting. RO and UF membrane elements for this service should be specified for high-pH and high-silica tolerance given the Kolkata concentrate profile.
| Configuration | Local RO recovery | Overall recovery (with salt precipitation) | Permeate TDS (mg/L) | Permeate silica (mg/L) | Concentrate handling |
|---|---|---|---|---|---|
| Conventional BWRO (no precipitation) | 75–80% | 75–80% | 10–50 | 1–5 | Large, scaling-prone brine to MVC/ZLD |
| Conservative first-pass RO | 50–70% | 50–70% | 10–50 | 1–5 | Routed to fluidized-bed crystallizer |
| RO + controlled salt precipitation + 2nd-pass RO (MAXH₂O logic) | 85–90% each pass | ~95% | <50 | ~1 | Salt pellets dewatered; residual NaCl brine minimal |
| Dynamic RO (cyclic production/flush) | Up to 90% single stage | Up to 90% | <50 | <2 | Smaller brine, longer CIP intervals |
Reuse, discharge, or ZLD: a 2026 decision framework for Kolkata
The default for a 2026 Kolkata hyperscale is reuse for cooling-tower makeup: it gives the highest water savings, the lowest OPEX, and it sidesteps most of the WBPCB discharge negotiation. The 50,000 GPD RO block that produces this makeup typically lands at $1.50–3.00 per thousand gallons treated (per Genesis Water Tech, 2026) — flag this as a global benchmark; India OPEX will track lower on energy and labor but higher on membrane logistics, so treat the band as directional rather than as a quote. Add an MVC or brine concentrator at 95–98% recovery on the RO concentrate (distillate <10 mg/L TDS) when reuse demand is satisfied but the concentrate still exceeds the WBPCB TDS cap. Full ZLD with a crystallizer becomes economic only when the consent imposes a near-zero-discharge condition or freshwater is capped; Genesis Water Tech (2026) places ZLD CAPEX at $3–8M and OPEX at $5–15/kgal — globally benchmarked and likely the upper bound for India, where EPC and energy deltas cut both ways. Partial ZLD (80–90% volume reduction) is the pragmatic middle ground when the concentrate can be sent to a downstream industrial user or a designated disposal hauler under a WBPCB special permit. For the chemistry-sensitive blocks — antiscalant, biocide, pH correction — use a dosing skid tied to the RO feed conductivity, and follow it with a high-efficiency sedimentation stage ahead of the membranes if the blended stream carries more than 30 mg/L TSS.
| End state | Indicative CAPEX | Indicative OPEX | Overall water recovery | WBPCB Consent complexity | Best-fit Kolkata scenario |
|---|---|---|---|---|---|
| Reuse as CTBD makeup (RO only) | Lowest | $1.50–3.00/kgal (global benchmark) | 50–85% | Low — internal reuse, no discharge consent for the stream | Default for new hyperscale; KMWC + Hooghly supply available |
| Reuse + MVC on concentrate (partial ZLD) | Moderate | Mid (driven by thermal energy) | 85–95% | Moderate — concentrate handling needs clarification | Sites with WBPCB TDS cap below concentrate TDS |
| Full ZLD with crystallizer | $3–8M (global benchmark, India likely lower) | $5–15/kgal (global benchmark) | 95–99% | High — no-discharge consent, more monitoring | Notified Ganga-basin no-discharge zones, capped freshwater intake |
| Discharge to municipal sewer (post-treatment) | Low–moderate | Discharge fees plus treatment OPEX | 0% (water leaves site) | High — site-specific consent, tighter than CPCB defaults | Avoid unless freshwater is abundant and consent is straightforward |
For the Indian context, treat the global CAPEX/OPEX bands as directional and overlay them with India-specific cost commentary — typically lower EPC and energy on one side, higher membrane logistics and RO antiscalant cost on the other — rather than quoting a specific INR figure that the SERP research does not support.
Sludge, monitoring, and the 2026 consent checklist
Solids handling and online monitoring are the parts of a CTBD reuse plant that engineers routinely under-resource. Salt pellets from the fluidized-bed reactor, clarifier sludge, and any RO concentrate solids are dewatered on a plate-and-frame filter press with the filtrate returned upstream — the cake goes to TSDF, the water stays in the loop. Online instrumentation should at minimum cover conductivity, pH, ORP, turbidity, and silica at the RO feed and permeate, with a UV stage downstream of the polishing step if humidification makeup is in scope; the data logger feeds the WBPCB compliance file directly. Submit the Consent to Operate application before the RO train is sized, not after — the consent conditions, not the equipment list, define the design envelope. The quarterly effluent monitoring program under a typical WBPCB CTO covers flow, pH, TDS, TSS, BOD, COD, oil & grease, and residual chlorine; plan sampling points at the equalization tank outlet, RO permeate, and final discharge (or reuse) header, with a dedicated sampling tap for the SPCB officer. For pH boundary conditions across Indian states, the 2026 pH discharge limit guide is a useful cross-check against the consent schedule.
Frequently Asked Questions
Does a data center in Kolkata need Zero Liquid Discharge in 2026?
Not by default. Full ZLD is imposed only when WBPCB consent flags the site as a no-discharge zone or caps freshwater intake; the typical 2026 hyperscale in Salt Lake Sector V or Rajarhat is cleared on reuse-plus-partial-ZLD, where the RO concentrate is reduced by 80–90% via MVC and the residual brine is hauled or sent to a designated industrial user. A consent-driven ZLD crystallizer adds $3–8M in global-benchmark CAPEX (per Genesis Water Tech, 2026), so it should be specified only after the consent text is in hand.
What is the realistic recovery rate for cooling tower blowdown RO in Kolkata?
Conventional BWRO on CTBD plateaus at 75–80% recovery before silica and calcium scaling force shutdown; with controlled salt precipitation and a two-pass configuration, overall recovery reaches ~95% with permeate silica of ~1 mg/L (per IDE-Tech, 2026). A first-pass design that does not budget for salt removal should not be quoted above 70% recovery in the consent application.
How does WBPCB Consent to Operate differ from CPCB Schedule-VI for a hyperscale data center?
CPCB Schedule-VI effluent tolerances and Schedule-I general standards set the national floor, but WBPCB writes site-specific conditions — typically a TDS cap in the 2,000–2,100 mg/L range and BOD below 30 mg/L for sewer discharge — that are tighter than the national defaults. The CTO application must include the proposed discharge points, a quality monitoring plan, and the reuse end-use declaration, and it should be filed before the RO train is sized so the design tracks the consent, not the other way around.
What is the smallest equipment set a Kolkata data center should specify to make CTBD reuse bankable?
For a 2026 hyperscale, the minimum defensible train is side-stream filtration (1–5% of circulation, 10–25 µm self-cleaning screens), a UF pretreatment ahead of RO, a high-recovery industrial RO system with dynamic operation, an antiscalant/biocide dosing skid, and a plate-and-frame filter press for the salt pellets. Anything less leaves either the membranes or the consent exposed. For an Indian supply-and-spec conversation, route the sizing through HydropureWater's engineering team with the Hooghly intake TDS swing and the WBPCB consent text attached.
