Why Almaty Data Centers Must Treat Cooling Tower Blowdown as a Resource, Not Waste
A data center in Almaty, Kazakhstan needs a treatment train that handles cooling tower blowdown (CTBD) — high in TDS, silica, calcium hardness and residual biocides — plus glycol-bearing closed-loop drains. The standard 2026 scope is pretreatment (multi-media filter + cartridge), antiscalant dosing, brackish RO at 75–80% recovery, and a polishing/disinfection step, with a target of 4–6 cycles of concentration. Blowdown volume equals roughly 1/(CoC−1) of makeup water, so moving from 4 to 6 CoC cuts blowdown by about 5 percentage points and reuse-eligible water by tens of thousands of liters per day per MW.
Almaty sits in the Ili–Almaty basin, where the Kapchagay reservoir and the snowmelt-fed rivers of the northern Tien Shan supply a city of roughly two million people, downstream irrigation in the Almaty and Zhetysu regions, and growing industrial demand. Kazakhstan's Water Code and the 2024–2026 municipal water-supply strain documented by the Almaty region administration have already pushed the city toward metered allocation during summer low-flow months. For a 10–20 MW colocation site — the size that dominates the current Almaty pipeline — that means competing with households and agriculture for the same limited surface-water intake, not an abstraction.
The scale is real. A 100 MW facility can withdraw up to 2,000,000 L/day, per ide-tech's 2026 reference (per ide-tech, 2026). Scale that to a 10 MW Almaty site and you sit at roughly 200,000–400,000 L/day of intake; at 4 CoC, the blowdown is 1/(4−1) ≈ 25% of makeup, or 50,000–100,000 L/day of recoverable CTBD per 10 MW. Genesis Water's 2026 analysis puts industry WUE at 1.8–2.5 L/kWh (0.47–0.65 gal/kWh), and the U.S. DOE claims a 3→6 CoC shift cuts makeup by about 20% and blowdown by about 50% (per DOE cooling-tower guidance). That "50%" claim is overstated for the 4→6 step, which actually moves blowdown from 25% to 20% of makeup — a 5 percentage-point swing, or a 20% relative reduction, not a 50% cut. The distinction matters when a Kazakh client asks what the real water saving is.
There is also a definitional trap that affects permitting. Water consumption is the volume permanently removed from the watershed through evaporation; water usage is the broader intake figure that includes water withdrawn, used, and returned (per Genesis Water, 2026). A facility can report a flattering WUE while still discharging large volumes of contaminated blowdown. Treating CTBD as an internal resource — not as a compliance headache — is the only way to lower both numbers simultaneously, which is what the Almaty region's water office is increasingly asking for in new site assessments.
What CTBD and Data-Center Wastewater Streams Actually Contain
CTBD is not generic "industrial wastewater." It is already-conditioned cooling water, heated to useful temperature and dosed with the chemicals needed to keep a cooling tower alive. Typical chemistry, per HOBI's 2026 technical reference: TDS 500–2,500 mg/L, hardness 400–1,200 mg/L as CaCO₃, silica 20–80 mg/L, suspended solids 20–200 mg/L, residual free chlorine 0.1–1.0 mg/L, plus isothiazolinone biocides, phosphonate scale inhibitors, and trace metals (Cu, Fe, Zn) from corrosion (per HOBI, 2026). In Almaty, the feedwater is snowmelt-fed surface water with seasonal turbidity swings — Q2 spring runoff can push raw-water turbidity above 50 NTU, and winter operation concentrates calcium hardness to its annual peak. Jar-test verification on at least two seasonal samples is non-negotiable before you lock in the antiscalant dose.
Ancillary streams complicate the picture. Propylene glycol–water closed loops, used for freeze protection in outdoor piping, dry coolers, and some liquid-cooling secondary loops, drain during maintenance or upset events. Propylene glycol is preferred over ethylene glycol because its aquatic toxicity is lower (per HOBI, 2026), but at 30–50% concentration it still imposes a high BOD load and must be segregated from the CTBD stream rather than blended in. Humidification bleed-off, generator cooling-water dumps, and fire-system test water round out the inventory; the last is normally once or twice a year but can be 50–200 m³ in a single event.
Two terms you will see in every Kazakh permit submission, defined here for the project file:
- Cycles of Concentration (CoC) — the ratio of dissolved solids in the circulating cooling water to dissolved solids in the makeup water. A system at 4 CoC concentrates minerals four-fold before blowdown (per Genesis Water, 2026).
- Water Usage Effectiveness (WUE) — annual site water usage divided by IT equipment energy, in L/kWh (or gal/kWh). The Green Grid metric that the DOE references for federal data-center reporting (per HOBI, 2026).
| Parameter | Typical CTBD range | Design target after pretreatment | Almaty-specific note |
|---|---|---|---|
| TDS | 500–2,500 mg/L | < 1,000 mg/L to feed BWRO | Winter peak due to lower basin inflow |
| Total hardness as CaCO₃ | 400–1,200 mg/L | < 200 mg/L post-softener or with antiscalant | Ca²⁺ dominant in Ili basin water |
| Silica (SiO₂) | 20–80 mg/L | < 30 mg/L feed to RO at 75% recovery | Limits RO recovery ceiling |
| Free chlorine | 0.1–1.0 mg/L | < 0.1 mg/L (dechlor before RO) | Sodium bisulfite dosing standard |
| TSS | 20–200 mg/L | < 10 mg/L post-DAF/MMF | Spring turbidity spikes in feed |
| Propylene glycol (closed loop) | 0–50% (event-driven) | Segregated, treated via MBR | Winter drain events, −25 to −30 °C |
Kazakhstan and Almaty Discharge Rules a Data Center Has to Meet
Kazakhstan runs a two-track discharge regime. The first track is a negotiated connection to the AlmatyKogamChygaru (Almaty wastewater utility) sanitary sewer, where the utility acts as the pretreatment control authority and sets site-specific limits on flow, pH, temperature, suspended solids, oil and grease, residual chlorine, and sulfate. The second track is direct discharge to surface water under the Water Code and SanPiN 3.01.067-97, with general limits of pH 6.5–8.5, suspended solids ≤ 50 mg/L, oil/grease ≤ 0.3 mg/L, residual chlorine ≤ 1.5 mg/L, and BOD₅ plus sulfate caps that vary by receiving body. The exact 2026 fee schedule, in KZT/m³, is set jointly by the Department of Ecology and AlmatyKogamChygaru and changes annually; confirm the current figures with both before the design basis is frozen.
Non-compliance penalties under the Ecological Code include administrative shutdown orders, fines scaled to environmental damage, and revocation of the discharge permit — all of which become public-record events that hyperscale tenants will treat as reputational risk. The pattern is familiar: when operators are forced to disclose blowdown chemistry, as Amazon was after the South Bend disclosure (per South Bend mayoral action reported in 2026), the downstream scrutiny multiplies. Glycol streams complicate this further. In Kazakhstan, glycol-bearing drainage is generally classified as a hazardous waste, must be segregated at source, and cannot be blended into the CTBD line for discharge; doing so puts the operator outside the sewer permit and into a hazardous-waste handler regime.
For the EPC engineer, the practical implication is that the design must support two separate terminal points: a sanitary-sewer connection for treated CTBD, and a licensed waste-hauler route for spent glycol. Treated permeate destined for reuse as cooling-tower makeup avoids both discharge paths entirely, which is why reuse is the most defensible option in front of a Kazakh regulator.
Treatment Train for Almaty CTBD: Pretreatment, RO, Reuse, and Brine Handling
The standard 2026 CTBD treatment train for a 5–20 MW Almaty site is a six-step modular skid, designed around brackish RO at 75–80% recovery as the workhorse. Below is a step-by-step walkthrough with the engineering numbers you need to defend the equipment list.
Step 1 — Equalization and screening. A rotary bar screen (2–5 mm aperture) removes leaves, ice, and debris from the cooling-tower basin and humidification bleed. An equalization tank sized for 8–24 h residence — for a 10 MW site producing 50–100 m³/day of CTBD, that is 17–100 m³ of buffer — smooths the diurnal swing and lets downstream units run at steady state. Remember the blowdown math: blowdown volume = makeup / (CoC − 1).
Step 2 — Pretreatment. A DAF unit for TSS and FOG removal from CTBD removes oil, grease, and a portion of the suspended solids that come off the tower fill, typically bringing TSS below 30 mg/L. The water then passes through a multi-media filter to bring CTBD SDI below RO limits — sand, anthracite, and garnet layered for 10–15 micron effective capture. Target SDI₁₅ under 5 at the RO feed. A automatic antiscalant and pH-adjustment dosing skid injects a phosphonate- or polymeric-antiscalant tailored to the silica and calcium levels in the equalized feed.
Step 3 — Cartridge filtration. A 5 µm cartridge guard filter protects the RO membranes from any media-filter carryover. Differential-pressure instrumentation with automatic shut-in at 0.7 bar prevents membrane exposure to plugged cartridges.
Step 4 — Brackish reverse osmosis. An industrial RO system sized for Almaty CTBD reuse runs at 75–80% recovery — the conventional ceiling before silica, calcium carbonate, and calcium sulfate scaling become unmanageable (per ide-tech, 2026). Concentrate is either routed to a brine-handling step or sent to a small lined evaporation pond, with a high-recovery (90–95%) desalter using controlled salt precipitation on seed material as a justified upgrade for sites where freshwater cost or sewer fees dominate the OPEX.
Step 5 — Permeate polishing and reuse. A UV sterilizer for permeate disinfection in a closed reuse loop, or a chlorine dioxide generator for biological control, polishes the permeate before it returns to the cooling-tower makeup line, or branches to irrigation or toilet flushing if those reuse pathways are approved. The same UV train protects the closed reuse loop from Legionella and biofilm carryover, which is the most common cause of post-installation microbiological failures (HydropureWater field data, 2026).
Step 6 — Glycol stream. Propylene glycol drains are routed to a biological MBR (moving-bed biofilm reactor) sized for a 30–50% glycol feed, with optional ozone polishing for residual organics before the treated effluent is blended with the CTBD line or discharged under hazardous-waste manifests.
| Step | Unit operation | Key parameter | Typical range / target |
|---|---|---|---|
| 1 | Bar screen + equalization | Residence time | 8–24 h |
| 2 | DAF + multi-media filter | SDI₁₅ to RO | < 5 |
| 3 | 5 µm cartridge | ΔP shut-off | 0.7 bar |
| 4 | BWRO | Recovery | 75–80% (up to 95% with desalter) |
| 5 | UV / ClO₂ | UV dose | ≥ 40 mJ/cm² |
| 6 | MBR for glycol | Glycol feed | 30–50% (event-driven) |
Sizing, Energy, and Cost Reality for a 10 MW Almaty Site
A worked example: a 10 MW evaporatively cooled Almaty colocation site at 4 CoC, makeup 300 m³/day, blowdown ≈ 1/(4−1) × 300 = 100 m³/day. Designing RO at 80% recovery yields 80 m³/day of reusable permeate and 20 m³/day of concentrate — concentrate that goes either to a brine-recovery desalter or, more commonly at this scale, to a small lined evaporation pond or to the AlmatyKogamChygaru sewer under negotiated limits. Energy for the RO train sits at roughly 0.6–1.2 kWh/m³ of permeate, plus feed and booster pumping; for a 60–80 m³/day permeate stream, total loop electricity is on the order of 50–80 kWh/day, a minor load against a 10 MW IT envelope.
CAPEX for a modular 10–20 m³/h blowdown treatment skid — DAF, multi-media filter, dosing, cartridge, BWRO, UV — typically lands in the USD 250,000–450,000 range (HydropureWater field data, 2026), with the RO membrane replacement as the largest single consumable at roughly 25–30% of annual OPEX. KZT water tariffs in Almaty and the AlmatyKogamChygaru sewer fee together are the swing variables: when combined with avoided potable makeup, the simple payback band sits at 3–5 years, consistent with Genesis Water's 2026 framework for a 15 MW reference case. Tariff sensitivity in KZT is the single biggest reason a small reduction in CoC-related OPEX matters more than a more efficient RO pump.
Frequently Asked Questions
What is the actual permit process for discharging treated CTBD to the AlmatyKogamChygaru sewer?
The operator submits a pretreatment application with influent/effluent characterization to AlmatyKogamChygaru, negotiates site-specific limits on flow, pH, temperature, suspended solids, oil/grease, residual chlorine, and sulfate, and signs a discharge agreement with KZT-anchored fees. Final limits and 2026 fees must be confirmed with the utility and the Department of Ecology before the design basis is frozen.
How should propylene glycol from closed-loop drains be handled in an Almaty winter?
Segregate glycol streams at source. Route to a biological MBR sized for 30–50% glycol feed, with optional ozone polishing. Never blend into the CTBD line for sewer discharge — under Kazakh classification the spent glycol is a hazardous waste and must travel via licensed hauler. Containment and leak detection on outdoor propylene glycol loops are a basic expectation in −25 to −30 °C Almaty winters.
What is the realistic recovery ceiling for a brackish RO unit on Almaty CTBD?
Conventional BWRO holds 75–80% recovery before silica, calcium carbonate, and calcium sulfate scaling force a clean-in-place cycle. A MAXH₂O-style brine desalter using controlled salt precipitation on seed material can push overall system recovery to 90–95% with silica in the permeate around 1 mg/L, but only when the KZT economics of the AlmatyKogamChygaru sewer fee and water tariff justify the added CAPEX (per ide-tech, 2026).
What do hyperscale operators do differently from a 5–20 MW Almaty colocation site?
Hyperscale sites (100+ MW) can justify multi-stage RO, ion-exchange polishing, and full-time specialist operators because their CAPEX per m³ treated falls by 3–4× compared to smaller facilities (per Genesis Water, 2026). For a 5–20 MW Almaty colocation, the right answer is a modular skid, standard BWRO at 75–80% recovery, and reuse as cooling-tower makeup — anything more elaborate usually sits idle or operates inefficiently.
