Why Yerevan Forces a Self-Contained CTBD Train in 2026
A hyperscale data center in Yerevan in 2026 needs a six-stage on-site train — segregation/equalization, DAF plus multi-media filtration, softening with antiscalant dosing, MBR polishing, two-pass RO at 80-95% recovery, and ZLD on the RO brine only — designed to the Hrazdan River basin TDS baseline and the Ministry of Environment permit framework under the RA Water Code, because the municipal sewer is sized for domestic waste and the Ararat basin is allocation-capped.
IDE-Tech's 2026 reference figure of 2,000,000 L/day for a 100 MW facility is the starting point for any Yerevan feasibility study. Translated to the Ararat basin, that volume collides with a hydrological reality the basin's surface and groundwater withdrawal is already at or above the 1990s-era cap tied to Lake Sevan policy, and the State Committee of Water Systems has no new allocation headroom in 2026 (per the RA Water Code, 2002, as amended). The implication is direct: any hyperscale withdrawal beyond what the site can re-use internally is not a commercial question, it is a permit pre-condition that is already closed.
The Yerevan municipal sewer compounds the constraint. The Yerevan Jur plant is sized for roughly 500,000-600,000 m3/day of domestic sewage, with no industrial CTBD load class. A 5-20 MW hall generating 200-1,000 m3/day of blowdown collides with a network designed for BOD, not brackish recirculated cooling water. Discharge to municipal sewer in 2026 is not a design option for hyperscale flows, because the interceptor reaches and the receiving water (the Hrazdan through Yerevan) carry summer TDS 400-700 mg/L and winter TDS up to 900 mg/L in low-flow years — a baseline that fixes the de facto effluent TDS ceiling for any negotiation with the Ministry of Environment (MoE). The legal anchors for that negotiation are the RA Water Code and the 2014 Law on Environmental Impact Assessment (amended 2021), and any new data-center project that triggers a meaningful Ararat aquifer withdrawal is treated as politically infeasible — so reuse becomes a permit precondition, not an economic choice. For peer context on the broader regional practice, the peer Erbil data center guide covers the KRG version of the same constraint.
The CTBD Chemistry a Yerevan Hyperscale Hall Must Handle
CTBD chemistry sets the envelope. Per Ecologix's 2026 reference profile, the stream leaving a 4-6 COC tower carries TDS up to 2,000 ppm, effluent temperature 30-40°C, Cu/Zn/Fe corrosion products, isothiazolinone biocide residuals, phosphonate antiscalants, and biofilm suspended solids. Cycles of concentration (COC) at 4-6 are the operating band, and the blowdown formula B = E/(COC-1) converts evaporation rate E to blowdown volume for any heat-load assumption — at 4 COC, blowdown equals 25% of make-up; at 6 COC, 20%. The three sparingly soluble salts that drive scaling risk are silica, calcium carbonate, and calcium sulfate, and conventional BWRO plateaus at 75-80% recovery before scaling becomes unmanageable (IDE-Tech, 2026).
Yerevan-specific make-up water matters. The Ararat aquifer baseline runs 300-500 mg/L TDS, moderately hard, with seasonal silica variation; a direct Hrazdan intake in summer pushes higher. Langelier Saturation Index (LSI) is held between -0.5 and +0.5, and Ryznar Stability Index (RSI) above 6 for non-aggressive operation — these targets feed the antiscalant selection downstream. Cooling-tower make-up quality targets the engineer should enforce: TDS below 200 mg/L, hardness below 50 mg/L as CaCO3, silica below 10 mg/L as SiO2, chloride below 100 mg/L.
| Parameter | CTBD Influent (Yerevan, 4-6 COC) | Cooling-tower make-up target | Notes |
|---|---|---|---|
| TDS | up to 2,000 ppm | < 200 mg/L | Hrazdan summer baseline 400-700 mg/L sets ambient |
| Temperature | 30-40 °C | ΔT < 5 °C above ambient at discharge | Drives MoE thermal envelope |
| Hardness as CaCO3 | 600-1,200 mg/L (COC-concentrated) | < 50 mg/L | Ararat make-up moderately hard |
| Silica (SiO2) | 40-80 mg/L (COC-concentrated) | < 10 mg/L | Drives second-pass recovery ceiling |
| Cu / Fe / Zn | 0.1-2 mg/L combined | < 0.1 mg/L each to RO | Corrosion products foul membranes quickly |
| Biocide (isothiazolinone) | 1-10 mg/L active | Below RO tolerance | Route through carbon or ClO2 side-loop first |
| LSI / RSI | +1.0 to +2.0 / 4.5-5.5 | -0.5 to +0.5 / > 6 | Antiscalant tied to feed flow |
The Six-Stage On-Site Train for a Stand-Alone Yerevan Data Hall
Stage 1 — Segregation and equalization. A dedicated EQ tank with 4-8 hour hydraulic retention and online pH/conductivity dampens the 1-5 pH excursions that follow chiller trips, which are routine on a hyperscale site. Humidification and AHU condensate (typically < 50 mg/L TDS) is segregated into its own line because glycol from coil leaks requires stripping, not blending into the main RO loop.
Stage 2 — DAF and multi-media filtration. A ZSQ DAF system in the 4-300 m³/h class floats oils, biofilm, and metal-hydroxide floc, followed by a multi-media filter that drops SDI below 3 and protects the RO from Cu, Fe, and Zn fouling. This combination is the workhorse for the corrosion-product load in CTBD.
Stage 3 — Softening and antiscalant dosing. A twin-tank water softener (KJ-WT series, 1-45 T/h class) targets hardness below 50 mg/L as CaCO3 and silica below 10 mg/L, with a PLC-controlled antiscalant dosing skid tied to RO feed flow handling residual scaling potential that softening cannot reach.
Stage 4 — MBR polishing (optional for stand-alone). A submerged PVDF MBR polishing stage with 0.1 µm membranes delivers < 1 NTU and < 10 mg/L COD, allowing direct RO feed without media filtration. For a stand-alone Yerevan data hall with no co-located fab, MBR is optional unless sanitary load is co-mingled — the Luanda data center guide covers the same logic for a different receiving basin.
Stage 5 — Two-pass RO at 80-95% recovery. An industrial RO unit delivers permeate at < 50 mg/L TDS. First pass runs at 150-250 psi (10-17 bar) for bulk salts; second pass polishes to cooling-tower make-up spec — TDS < 200 mg/L, hardness < 50 mg/L, Cl⁻ < 100 mg/L. Recovery tuning is per stream, not a single number, and 80-95% is the practical operating window — above 95% silica scaling on second-pass membranes drives CIP frequency up sharply.
Stage 6 — ZLD on RO brine only. Mechanical vapor recompression at 25-40 kWh/m³ of brine concentrated; full-stream ZLD is over-specified for a stand-alone data hall, reserve it for the RO brine when the Hrazdan discharge path is restricted. Side-stream filtration at 1-5% of total circulation flow, using 10-25 µm self-cleaning spiral units, drops suspended solids in blowdown to levels the RO can handle without pre-coat — the cheapest way to push cycles of concentration higher and reduce what the RO train has to do.
| Stage | Equipment | Hand-off water quality | Design notes |
|---|---|---|---|
| 1 — Segregation / EQ | EQ tank, 4-8 h HRT, online pH/Cond | pH 6.5-8.5, flow damped | Condensate segregated (< 50 mg/L TDS) |
| 2 — DAF + MMF | ZSQ DAF 4-300 m³/h, multi-media filter | SDI < 3, TSS < 5 mg/L | Cu/Fe/Zn stripped as floc |
| 3 — Softening + antiscalant | KJ-WT softener + dosing skid | Hardness < 50 mg/L CaCO3, SiO2 < 10 mg/L | LSI -0.5 to +0.5, RSI > 6 |
| 4 — MBR (optional) | PVDF submerged, 0.1 µm | < 1 NTU, < 10 mg/L COD | Only if sanitary co-mingled |
| 5 — Two-pass RO | Industrial RO, 150-250 psi / 10-17 bar | TDS < 200 mg/L, Cl⁻ < 100 mg/L | 80-95% recovery per pass |
| 6 — ZLD on brine | MVC 25-40 kWh/m³ | NaCl-rich solid waste | Only on RO brine, not full stream |
Permit Pathway: Negotiating with the Ministry of Environment in 2026
The statutory stack is the RA Water Code (2002, amended), the Law on Environmental Impact Assessment (2014, amended 2021), Government Decree N 59-N on effluent limits, and any 2026 MoE guidance specific to data centers. Hyperscale flows push a project into full EIA rather than screening, because the freshwater withdrawal threshold and the discharge envelope both trip the higher-tier review. The negotiating counterparty is the MoE Environmental Impact Assessment Expert Center.
The typical MoE limit band for 2026 industrial discharge to the Hrazdan is TDS < 1,000 mg/L, temperature ΔT < 5 °C, heavy metals to trace mg/L — anchored to the Hrazdan receiving-water baseline of summer TDS 400-700 mg/L. The closed-loop / > 80% reuse expectation emerging in MoE practice for high-water-stress ICT projects is policy alignment, not generosity: projects that cannot demonstrate internal reuse above 80% find permit issuance slow and conditional. The IDE-Tech case data point — silica permeate ~ 1 mg/L at 95% recovery — is the number that lets a designer argue for a 95% recovery permit limit rather than the 75-80% default, and that argument lands in 2026 MoE practice because it directly addresses the Ararat basin allocation closure.
CAPEX, OPEX, and the Avoided-Discharge Payback in Yerevan
2026 CAPEX bands are engineering estimates, not firm quotes: a small data hall package plant plus haul-off runs ~$150-300 per m³/day installed; mid-size 200-1,000 m³/day MBR + RO runs ~$400-700; hyperscale with ZLD on brine runs ~$800-1,200. A 5-20 MW Yerevan hall typically sits in the mid-size band. The avoided-discharge math at $5-15/kgal makes the train fund itself: 100 m³/day of untreated blowdown at the upper end is ~USD 400/day, so an 80% recovery RO typically pays back inside 24 months at hyperscale flow (HydropureWater field data, 2026).
The line items that push OPEX up are membrane CIP frequency above 95% recovery due to silica scaling on second-pass, biocide residuals shortening RO life (the case for a chlorine dioxide generator upstream of the RO rather than isothiazolinone in the tower), and MVC electricity at 25-40 kWh/m³ of brine. The Erbil guide established the rule that holds here too: below ~300 m³/day the OPEX gap between RO-only and RO+ZLD makes brine haul-off cheaper than MVC, unless the site is in a zero-discharge zone along the Hrazdan. For a peer reference, the Rosario data center guide covers a comparable ZLD decision under a different river-basin constraint.
| Scope | CAPEX (USD per m³/day installed) | Notes |
|---|---|---|
| Small data hall, package plant + haul-off | $150-300 | < 200 m³/day; brine removed by tanker |
| Mid-size 200-1,000 m³/day, MBR + RO | $400-700 | Typical 5-20 MW Yerevan hall |
| Hyperscale, ZLD on brine | $800-1,200 | Only if Hrazdan discharge restricted |
| Avoided-discharge payback at $5-15/kgal | ~ 24 months at 100 m³/day, 80% recovery | Funds the train internally |
Frequently Asked Questions
What is the minimum recovery rate a Yerevan data center should target on its CTBD train?
80% is the floor that makes the avoided-discharge economics close, and 95% is the practical ceiling before silica scaling on second-pass membranes drives CIP frequency up sharply. The operating window for a 2026 Yerevan hall is 80-95% recovery on the two-pass RO, with first-pass tuned per stream rather than as a single number.
Does a stand-alone data hall in Yerevan need ZLD, or is RO brine discharge acceptable?
A stand-alone data hall with no co-located fab runs MBR + two-pass RO at 80-95% recovery as the baseline; ZLD on RO brine via MVC at 25-40 kWh/m³ is reserved for sites where the Hrazdan discharge path is restricted or the project sits in a zero-discharge zone. Below ~300 m³/day, brine haul-off is usually cheaper than MVC.
Which Armenian regulator signs off on the discharge permit, and what is the typical review timeline in 2026?
The Ministry of Environment (MoE) Environmental Impact Assessment Expert Center is the negotiating counterparty under the RA Water Code and the Law on Environmental Impact Assessment. Hyperscale flows push a project into full EIA rather than screening, and the 2026 review timeline for full EIA typically runs 90-180 days from submission, conditional on the project demonstrating > 80% internal reuse.
How does the Ararat basin water allocation policy affect a new hyperscale project in Yerevan?
The Ararat basin surface and groundwater withdrawal is already at or above the 1990s-era cap tied to Lake Sevan policy, and any new hyperscale withdrawal from the Ararat aquifer is treated as politically infeasible in 2026. Reuse above 80% is therefore a permit precondition, not an economic option, and the train must be designed for closed-loop operation from day one.
What is the realistic payback period for a six-stage MBR + two-pass RO train at 80-95% recovery?
At 100 m³/day of blowdown and avoided-discharge cost of $5-15/kgal, an 80% recovery RO pays back inside 24 months at hyperscale flow (HydropureWater field data, 2026). The 95% recovery ceiling extends payback further only when the Hrazdan discharge path is restricted; otherwise the OPEX penalty from silica CIP frequency erodes the gain.
