Why Seoul is Different: Water Stress, Permitting, and the 2026 Baseline
A Seoul data center in 2026 cannot treat cooling-tower blowdown as a routine discharge because the Han River basin sits under formal water-allocation review, and the Korea Water Pollution Control Act (수질 및 수생태계 보전에 관한 법률) sets a national effluent envelope that a 30-50 MW hall will brush against the first month it operates. A 100 MW hyperscale site can draw up to 2 million liters of water per day (source: IDE-tech, 2026), and a 30-50 MW Seoul hall proportionally falls in the 600,000-1,000,000 L/day range. Seoul Waterworks (서울시 상수도사업본부) no longer treats the connection as routine and routes the application through a non-routine industrial-use review, including a documented water-stewardship plan. The engineering target is therefore not total facility flow but the blowdown fraction: 20-40% of intake leaves the cooling loop as purge (source: Genesis Water, 2026), which is currently discharged to the sewer. Industrial electricity from KEPCO at roughly ₩100-160/kWh is the OPEX anchor that determines whether high-recovery RO on that 20-40% pays back in a Korean context; that price is now firmly above the threshold where membrane reuse beats incremental freshwater plus sewer charges.
Cooling Tower Blowdown Chemistry in Seoul Tap Water
Seoul municipal water, drawn from the Han River and treated at the Gwanghui, Amsa, and Jayang plants, falls within a moderate chemistry envelope that vendors accustomed to U.S. Southwest influents often misread. Total hardness runs 40-80 mg/L as CaCO₃, seasonal silica sits at 5-15 mg/L, chloride is low (typically <20 mg/L), and pH is neutral to slightly alkaline at 7.0-7.6 (HydropureWater field data, 2026). The blowdown ratio is 1/(CoC-1): 25% at 4 cycles, 20% at 6 cycles, 17% at 7 cycles (source: Genesis Water, 2026), and the dissolved-solids concentration effect is the basis for the cycles-of-concentration (CoC) framing in standard vendor literature. The failure modes that the EPC must design against are calcium carbonate scaling above 5-6 CoC, calcium sulfate precipitation once sulfate concentrates 4-5×, microbiologically influenced corrosion in the June-September humid season, and chloride pitting on stainless heat-exchange surfaces if cycles are pushed too aggressively without makeup softening. A multi-media filter polishing the RO feed is the standard first step against suspended-solids carry-through from the tower basin.
The 2026 Compliance Stack: WPCA, Han River TMDL, and Seoul Sewer Ordinances
The compliance stack a Korean EPC must clear is layered, and the on-site reuse train is necessary to keep the polishing-reject stream inside the regulatory envelope. The Korea Water Pollution Control Act (수질 및 수생태계 보전에 관한 법률) sets the national effluent limits for public-water discharges: pH 5.8-8.6, BOD ≤120 mg/L, COD ≤160 mg/L, SS ≤200 mg/L (per Korean MOEE effluent standards, 2025-12). The Sewerage Act (하수도법) Article 12, enforced through the Seoul Metropolitan Government sewer-acceptance ordinance, sets tighter site-specific caps on the polishing-reject stream that goes to municipal sewer — typically SS, BOD, and n-hexane extracts below the WPCA national envelope. Han River basin total maximum daily loads (TMDLs) cap total phosphorus and total nitrogen where the site falls inside a designated protection zone, which most Seoul industrial parcels along the river do. Analytical work runs to Korean Standard Methods (KS M ISO) equivalents of JIS K 0101/K 0102 — pH, SS, COD, T-N, T-P, n-hexane extracts, trace metals at the KS-specified detection limits — and that suite is what the on-site or contract lab must run to demonstrate compliance.
| Regulation | Parameter | Limit | Applies to |
|---|---|---|---|
| WPCA (수질 및 수생태계 보전에 관한 법률) | pH | 5.8-8.6 | Public-water discharges |
| WPCA | BOD | ≤ 120 mg/L | Public-water discharges |
| WPCA | COD | ≤ 160 mg/L | Public-water discharges |
| WPCA | SS | ≤ 200 mg/L | Public-water discharges |
| Sewerage Act Art. 12 / Seoul ordinance | SS, BOD, n-hexane | Site-specific, tighter than WPCA | Municipal sewer discharge |
| Han River basin TMDL | Total P, Total N | Site-specific caps in protected zones | Designated protection zones |
Equipment Train for a 30-50 MW Seoul Hall
The configuration below represents the P&ID requirements for a 30-50 MW evaporatively cooled hall on a Seoul industrial parcel. Each stage has a defined influent and effluent target so the contractor can size the unit operation, not the vendor. Sidestream filtration on the cooling loop — sand filter or cyclone separator — is sized to 5-10% of circulating flow and keeps suspended solids below the threshold that would otherwise force a blowdown event. Makeup softening on the freshwater leg lifts the achievable CoC before calcium-carbonate scaling becomes limiting. Coagulation and a DAF unit upstream of the RO membranes drop the Silt Density Index below 3 and turbidity below 0.5 NTU at the membrane feed, which is the standard threshold for brackish RO. A PLC-controlled anti-scalant and biocide dosing skid is sized to RO feed flow and tuned for silica and calcium sulfate inhibition. The high-recovery RO operates at ~80% local recovery with controlled salt precipitation and dynamic RO cycling to reach ~95% overall recovery, producing permeate with silica ≤1 mg/L, TDS <50 mg/L, and conductivity <100 µS/cm (target figures based on IDE-tech MAXH₂O benchmarks, 2026). The polished stream is disinfected with UV or on-site generated chlorine dioxide before return to the cooling tower as makeup, and sludge from the lamella clarifier underflow plus the RO concentrate solids is routed to a plate-and-frame filter press for the RO concentrate solids for dewatering to ~60% cake solids.
| Stage | Influent target | Effluent target | Note |
|---|---|---|---|
| Sidestream filtration (5-10% of circ. flow) | Cooling loop TSS | TSS < 30 mg/L | Reduces blowdown frequency |
| Makeup softening | 40-80 mg/L as CaCO₃ | < 10 mg/L as CaCO₃ | Lifts achievable CoC |
| Coagulation + DAF | Turbidity 5-20 NTU | Turbidity < 0.5 NTU | SDI reduction before RO |
| Anti-scalant / biodispersant dosing | RO feed | Per chemistry model | Silica + CaSO₄ inhibition |
| High-recovery RO (~80% local, ~95% overall) | SDI < 3, free Cl₂ < 0.1 mg/L | TDS < 50 mg/L, SiO₂ ≤ 1 mg/L, conductivity < 100 µS/cm | Permeate to cooling-tower makeup |
| UV or ClO₂ disinfection | Permeate | Microbial control | Before return as makeup |
| Sludge dewatering (plate-and-frame press) | Lamella + RO conc. solids | ~60% cake solids | Solid waste to handler |
Sizing the Reuse Train to Blowdown Flow, Not Facility Flow
Hyperscale economics do not translate 1:1 to a mid-size Seoul build, as per-gallon CAPEX at facilities under 5 MW runs 3-4× higher due to non-linear fixed engineering, controls, and membrane-housing costs (source: Genesis Water, 2026). The optimal 2026 configuration for a 30-50 MW Korean hall is a modular skid in the 100-300 GPM range sized to blowdown flow only, featuring a high-recovery industrial RO system paired with a UF pretreatment skid. As an illustrative reference case: a 15 MW water-stressed site spending roughly $200,000 of CAPEX to recover 60% of blowdown shows about a 6.7-year simple payback on water alone, compressing to 3-5 years once avoided freshwater connection fees, sewer discharge volume charges, and chemical savings are counted (source: Genesis Water, 2026). Translated to a KRW framing for a Seoul 30-50 MW hall at current industrial water and sewer tariffs, a ₩260 million CAPEX case targeting 60% recovery typically lands at a 3-5 year payback once all avoided-cost line items are included. Modular skid architecture also prevents commissioning delays when membrane elements or media require replacement.
Where the OPEX Actually Goes in Korea
KEPCO industrial electricity at roughly ₩100-160/kWh dominates RO OPEX in a Korean hall, and energy-recovery devices on the high-pressure stream cut that line item by 30-40%. Membrane replacement runs on a 3-5 year cycle for RO and 5-7 years for UF, and stocking a spares inventory at the on-site warehouse reduces swap downtime from days to hours (HydropureWater field data, 2026). UV lamp service for the polishing stream and chlorine dioxide generator consumables are recurring, along with the anti-scalant and biodispersant fed by the PLC-controlled dosing skid. Finance committees often overlook line items that improve the payback math: reduced Seoul Waterworks freshwater connection fees at the next permit cycle, lower sewer discharge volume charges once blowdown volume drops, and eligibility for KECO and MOTIE water-efficiency incentives once the documented reuse percentage clears the program threshold. The chlorine dioxide generator and UV sterilizer are the primary consumable-bearing units to budget against.
Frequently Asked Questions
What wastewater and cooling blowdown treatment does a data center in Seoul, South Korea need?
A 30-100 MW Seoul data center in 2026 needs a sidestream filtration + makeup softening + DAF + high-recovery RO (~95% overall) treatment train that returns 80-85% of cooling-tower blowdown as cooling-tower makeup, with a small polishing-reject stream discharged to the Seoul municipal sewer under Sewerage Act Article 12 caps (per Seoul Metropolitan Government ordinance, 2025-12).
Does a Seoul data center need a Zero Liquid Discharge system in 2026?
No. ZLD is not required under current Korea WPCA or Seoul sewer ordinances for most sites. The 2026 compliant path is high-recovery RO returning permeate to the cooling loop, with the polishing-reject stream discharged to the public sewer inside the Sewerage Act Article 12 envelope (per Sewerage Act 하수도법, 2025-12).
What is the realistic payback period for a blowdown reuse system in a Korean data center?
The payback period is 3-5 years once avoided Seoul Waterworks freshwater connection fees, sewer discharge volume charges, and chemical costs are counted, versus 6.7 years on water savings alone at the $200,000 / 15 MW reference scale (source: Genesis Water, 2026). In KRW terms, a ₩260 million CAPEX case targeting 60% blowdown recovery typically lands at 3-5 years in a Seoul water-stressed scenario.