Why Rabat Is a Special Case for Data Center Water
Rabat's climate is Mediterranean-to-semi-arid, with annual rainfall between roughly 200 mm and 300 mm in the Bouregreg basin — a figure that places the capital on the same hydrological stress tier as Phoenix or Northern Virginia, the two reference hotspots for hyperscale water risk (per Ecologix Environmental Systems, 2026). The basin allocation is already contested between municipal supply, agriculture, and industry, and ONEE potable water is the default makeup source for new builds. Cooling-tower blowdown (CTBD) released untreated to the RADEEMA sewer therefore competes with a public system that is itself operating at the edge of seasonal allocation.
This is not a hypothetical market. The Rabat-Technopolis DC1 facility (Tier-3) and the N+ONE cluster in Casablanca are live Moroccan anchor projects that confirm hyperscale and colocation developers are already siting capacity in this basin (per Data Center Catalog, 2026). Permit and sampling obligations sit under Loi 10-95 on water, Décret 2-14-499 on liquid discharges, and the NM standards (ISO 5667 sampling family) enforced by ONEE for supply and RADEEMA for sewer discharge. A "generic RO + DAF" design imported from a North American spec will overshoot on CAPEX, miss the ONEE/RADEEMA TDS and temperature envelope, and underestimate how much makeup a 10 MW site actually pulls.
Using the industry benchmark of Water Usage Effectiveness (WUE) = 1.8 L/kWh, even a 10 MW site running at PUE 1.2 generates 10/1.2 × 24 × 1,800 L = 360,000 L/day of IT-driven water demand, or roughly 432 m³/day once humidification and auxiliary loads are included (per Ecologix, 2026). That volume is small relative to Casablanca's urban demand but large relative to RADEEMA's industrial-discharge tolerance per site, which is what drives the treatment-train decision in the next sections.
How Much Cooling-Tower Blowdown a Rabat Facility Actually Generates
Sizing starts with evaporation. The standard relationship is E = (Heat Load × 860) / (ΔHvap × η). For a 50 MW IT heat load at η = 0.8, E ≈ 99,537 kg/h (per Ecologix, 2026). Blowdown then follows B = E / (COC − 1), and total makeup M = E + B + drift, with drift around 0.02% of circulation. A common procurement misconception is that doubling COC halves blowdown; in reality, moving from COC 4 to COC 6 cuts blowdown from 25% to 20% of makeup — a 5 percentage-point reduction, not 50% (per Genesis Water Technologies, 2026). Chemistry cost and biological risk rise non-linearly above 5–6 COC, so the honest lever is downstream treatment, not just COC creep.
For a typical 10 MW evaporatively cooled site at 4 COC, monthly makeup is in the order of 15 million gallons (≈ 57,000 m³) and recoverable blowdown is roughly 3.75 million gallons/month (≈ 14,200 m³/month), or about 470 m³/day. The chemistry of that stream — TDS near 2,000 ppm, temperature 30–40 °C, plus Cu, Zn, molybdate, and biocide residues — is what makes it a treatment problem rather than a dilution problem (per Ecologix, 2026).
| Site size (IT load) | Heat load (MW) | Evaporation E (m³/day) | Makeup M @ COC 4 (m³/day) | Blowdown B @ COC 4 (m³/day) | Makeup M @ COC 6 (m³/day) | Blowdown B @ COC 6 (m³/day) |
|---|---|---|---|---|---|---|
| 5 MW edge / colocation | ~5 | ~ 215 | ~ 286 | ~ 71 | ~ 258 | ~ 43 |
| 10 MW colocation | ~10 | ~ 430 | ~ 572 | ~ 143 | ~ 516 | ~ 86 |
| 50 MW hyperscale (Rabat-Technopolis-class) | ~50 | ~ 2,150 | ~ 2,860 | ~ 715 | ~ 2,580 | ~ 430 |
These are the m³/day values the P&ID and the RADEEMA permit application must both reference. At 50 MW, blowdown alone (≈ 430–715 m³/day) exceeds the entire IT-driven water demand of a 10 MW site, which is why the reuse decision is made at the front of the project, not after commissioning.
The Chemistry That Makes CTBD Hard to Reuse
CTBD is a brackish stream enriched in silica, CaCO₃, and CaSO₄ — the three sparingly soluble salts that define the recovery ceiling of any membrane system. As evaporation proceeds in the tower, these species concentrate well past their solubility limits; a conventional brackish water reverse osmosis (BWRO) train therefore plateaus at 75–80% recovery before scaling, antiscalant dose, and flux instability make operation uneconomic (per IDE-Tech, 2026). Pushing recovery further with extra RO stages and booster pumps is mechanically possible but commercially punishing at 5–10 MW scale.
Two indices govern scaling risk. The Langelier Saturation Index (LSI = pH − pHs) is held between −0.5 and +0.5 on the RO feed, and the Ryznar Stability Index (RSI = 2 pHs − pH) is monitored in parallel because RSI catches underpredicted corrosion cases that LSI can miss (per Ecologix, 2026). Antiscalant and pH trim are delivered through a dosing skid sized to the feed flow, not the permeate flow. Biofouling follows first-order kinetics with k = 0.1–0.5 h⁻¹, and sludge settleability is judged against an SVI optimum of 80–150 mL/g when a clarifier is in the line (per Ecologix, 2026).
It is also worth being precise about the metrics. Cycles of Concentration (COC = TDStower / TDSmakeup) measures how concentrated the circulating water has become. Water Usage Effectiveness (WUE = annual site water / IT energy, in L/kWh) measures site intake, not what is reused versus dumped. A facility can post an acceptable WUE of 1.8 L/kWh and still lose 20–40% of intake as contaminated blowdown, because WUE is blind to discharge quality and reuse fraction (per Genesis Water Technologies, 2026). For a Rabat design review, COC, LSI, and recovery % belong on the P&ID; WUE belongs on the sustainability report. Cavitation and pump-side hydraulics are covered separately in our pump cavitation troubleshooting guide.
A Rabat-Ready Treatment Train: Multimedia → UF → RO → Polishing
The train below is modular, sized for the blowdown flows in the previous section, and consistent with ONEE/RADEEMA discharge limits on TDS, temperature, and metals.
- Stage 1 — Multimedia pretreatment and 5–10 µm pre-screen. A multimedia filter with sand, garnet, and anthracite layers removes TSS and protects the membranes downstream. Target feed SDI is below 5 (per Ecologix, 2026); a multimedia pretreatment filter is the first item on the equipment list.
- Stage 2 — Ultrafiltration (UF) guard. 0.03 µm PVDF hollow-fiber UF with automatic backwash and air scour strips colloids, bacteria, and turbidity upsets that would otherwise foul the RO. A UF pretreatment skid stabilizes feed quality so the RO can run at design flux without surprise CIPs.
- Stage 3 — Antiscalant, pH trim, and BWRO. An antiscalant and pH dosing skid holds LSI in the −0.5 to +0.5 window ahead of an industrial RO system operating at 75–80% local recovery. The permeate is already suitable as cooling-tower makeup without further polishing.
- Stage 4 — Concentrate handling. For closed-loop reuse above 80% recovery, route the RO concentrate through a controlled-precipitation or closed-circuit RO stage. The logic is the same as IDE's MAXH₂O architecture: deactivate scaling inhibitors in a fluidized-bed reactor so silica, CaCO₃, and CaSO₄ precipitate as dense pellets that discharge as a solid waste stream rather than a liquid brine (per IDE-Tech, 2026). The remaining NaCl brine can then be sent to a crystallizer.
- Stage 5 — Polishing and disinfection. ClO₂ generation on the permeate leg handles residual biofilm without forming trihalomethanes, which would breach RADEEMA reuse quality for humidification. A UV polisher on the permeate leg is the safer non-oxidizing alternative where the chlorine residual must stay at zero.
- Optional Stage 6 — ZLD via crystallizer. Where RADEEMA discharge is unavailable or where brine haulage to the inland Salé-Temara corridor is uneconomic, a falling-film or fluidized-bed crystallizer (per Saltworks, 2026) closes the loop entirely at the cost of roughly 2.5–3.0× the operating energy of a RO-only train.
| Stage | Unit operation | Key spec / target | Rabat-specific note |
|---|---|---|---|
| 1 | Multimedia + 5–10 µm screen | SDI < 5 | Handles Bouregreg surface-water turbidity spikes in winter |
| 2 | UF (PVDF hollow-fiber) | 0.03 µm, automatic backwash | Buffers variable CTBD quality from the cooling tower |
| 3 | Antiscalant + pH trim + BWRO | 75–80% local recovery, LSI ±0.5 | Permeate reused as tower makeup |
| 4 | Concentrate handling (CCRO / precipitation) | ≈ 95% overall recovery | Solids out as pellets, not brine |
| 5 | ClO₂ or UV polish | No THM formation, zero Cl₂ residual | Meets RADEEMA reuse criteria |
| 6 (optional) | Crystallizer / ZLD | Zero liquid discharge | Used when sewer is unavailable |
Onsite Treatment vs. Sewer Discharge in Morocco
Onsite treatment becomes mandatory, not optional, when any of three conditions holds (per Ecologix, 2026): effluent TDS exceeds 2,000 ppm, ΔT to the receiving sewer exceeds 5 °C, or the local WWTP cannot accept the 1.14–1.70 ML/day range that a hyperscale AI facility generates. In Morocco, that decision is filtered through Loi 10-95 on water, Décret 2-14-499 on liquid discharges, and the ONEE/RADEEMA industrial-effluent technical specification. RADEEMA's sewer acceptance criteria cap temperature rise and TDS at the manhole, and the authorization is site-specific rather than blanket.
For a 10 MW Rabat site, blowdown in the 240–340 m³/day range is well below any RADEEMA WWTP hydraulic limit, so direct discharge is technically feasible — but it forfeits 20–40% of the water the site has already paid to treat to ONEE potable standard (per Genesis Water Technologies, 2026). The decision rule is therefore economic: if reuse % × avoided makeup cost × water-stress factor exceeds the CAPEX + OPEX of a dosing and RO pretreatment package, the project builds onsite treatment and recycles; otherwise, it optimizes chemistry, pushes COC to 5–6, and pays the RADEEMA discharge fee.
Permit timing in the Rabat-Salé-Kénitra corridor typically runs 60–120 days for an RADEEMA authorization once the dossier is complete, and Loi 10-95 sampling must follow the NM ISO 5667 family for any analysis that is submitted as part of the application.
Sizing and Cost Snapshot for a Rabat Build (5 / 10 / 50 MW)
CAPEX for an RO train in the data-center segment is commonly benchmarked at up to $1.0 per m³ of daily treatment capacity, before freight, installation, and Moroccan engineering markup (per Ecologix, 2026). For a North African build, add 15–25% for logistics,海关, and ONEE-side electrical interface work; the figures below are planning-grade, not bid-grade.
| Site | Blowdown @ COC 4 (m³/day) | RO train size (m³/day permeate) | Multimedia + UF footprint | Indicative RO skid CAPEX (USD) | Polishing + dosing CAPEX (USD) |
|---|---|---|---|---|---|
| 5 MW edge | ~ 71 | ~ 50 | 2 m² | 50,000 – 90,000 | 25,000 – 40,000 |
| 10 MW colocation | ~ 143 | ~ 100 | 3 m² | 100,000 – 160,000 | 40,000 – 70,000 |
| 50 MW hyperscale (Technopolis-class) | ~ 715 | ~ 500 | 12 m² | 500,000 – 750,000 | 180,000 – 280,000 |
OPEX is dominated by chemical, membrane replacement (≈ 15–20% per year for RO elements in a well-run plant), and RO high-pressure-pump energy at 0.7–1.2 kWh/m³ permeate. On the credit side, a 10 MW site that recovers 60% of blowdown (roughly 3 million gal/year, or ≈ 11,400 m³/year) avoids both the ONEE industrial-tariff makeup charge and the RADEEMA discharge fee. A simple-payback model that ignores water-stress risk lands at about 6.7 years for a 15 MW site on $200k of CAPEX; priced properly with avoided discharge, energy, and the cost of a hypothetical ONEE curtailment event, payback compresses to 3–5 years (per Genesis Water Technologies, 2026).
Two practical warnings for Rabat developers. First, hyperscale RO/IX skids designed for 100+ MW sites fail economically at 5–10 MW — per-kg-of-treatment CAPEX runs 3–4× higher and the operating complexity exceeds the available staff (per Genesis Water Technologies, 2026). Modular industrial RO systems and a multimedia pretreatment filter sized to the actual m³/day are the right answer for the Rabat edge/colocation segment. Second, where concentrate is routed to a solids stream, dewatering is typically handled by a plate-and-frame filter press rather than a centrifuge, to keep indoor noise and particulate within the Technopolis site's environmental management plan; lifecycle O&M decisions are covered in our performance-based O&M contracts guide.
Frequently Asked Questions
What is the minimum treatment train for a 10 MW data center in Rabat?
Multimedia filtration to SDI below 5, ultrafiltration at 0.03 µm, antiscalant and pH trim, and a brackish water reverse osmosis unit at 75–80% recovery, followed by ClO₂ or UV polishing on the permeate leg. For a Rabat site under Loi 10-95, this train keeps blowdown TDS and temperature within RADEEMA sewer limits while recycling roughly 100 m³/day as cooling-tower makeup.
How much cooling-tower blowdown does a 50 MW Rabat hyperscale facility actually discharge?
At a 50 MW IT heat load, evaporation is about 2,150 m³/day. At COC 4, blowdown is roughly 715 m³/day; at COC 6, it falls to about 430 m³/day. The chemistry — TDS near 2,000 ppm, temperature 30–40 °C, plus Cu, Zn, and biocide residues — is what makes the stream a treatment problem rather than a dilution problem (per Ecologix, 2026).
Which Moroccan permits apply to cooling-tower blowdown discharge in Rabat?
Loi 10-95 on water, Décret 2-14-499 on liquid discharges, and the ONEE/RADEEMA industrial-effluent technical specifications. Sampling for the permit dossier must follow the NM ISO 5667 family, and a RADEEMA sewer authorization typically takes 60–120 days once the file is complete. Live anchor projects such as Rabat-Technopolis DC1 and the N+ONE Casablanca cluster are the working reference points for the type of facility regulators now expect to see in the application.
Is it cheaper to discharge to RADEEMA or to install an onsite RO reuse train?
For a 10 MW site, direct discharge at 240–340 m³/day is technically feasible, but a 60% RO recovery system pays back in roughly 6.7 years on a $200,000 CAPEX basis, and in 3–5 years once avoided ONEE makeup, avoided RADEEMA discharge fees, and water-stress risk are priced in (per Genesis Water Technologies, 2026). The economic crossover sits at the 10–15 MW scale under Rabat's current industrial-tariff structure. For comparison, the Astana data center blowdown guide, the Almaty data center blowdown guide, and the Mombasa data center wastewater guide cover the same sizing logic for those markets.
Related Equipment
- industrial RO system — specifications, capacity range, and technical data
