Cooling System Economics Analysis
sCO2 vs Conventional Cooling for Hyperscale Data Centers
Document ID: CCNL-CSE-2024-01 | Version: 1.0 | Date: December 2024
Executive Summary
This analysis compares the capital and operating costs of different cooling technologies for a 100 MW hyperscale data center in Newfoundland and Labrador. The key finding is that ocean-based sCO2 cooling delivers 35-45% lower total cooling cost over a 20-year period compared to conventional air cooling.
Cooling Technology Comparison
Capital Expenditure (CAPEX) per MW IT Load
| Cooling Technology | CAPEX/MW | Relative Cost | Notes |
| Traditional Air Cooling | $7.0M | Baseline | Chillers, CRAHs, containment |
| Direct Liquid Cooling | $6.3-6.5M | -7 to -10% | At 2x rack density (20kW/rack) |
| Immersion Cooling | $4.5-5.5M | -20 to -35% | Single-phase immersion |
| Seawater Air Conditioning (SWAC) | $4.0-6.0M | -15 to -45% | Depends on pipe distance |
| Ocean sCO2 Closed-Loop | $5.0-7.0M | -0 to -30% | Novel system, estimate |
NL sCO2 System CAPEX Breakdown (100 MW Facility)
| Component | Cost Range | Notes |
| Subsea HDPE Pipeline (8-10 km) | $8-15M | $1,000-1,500/m installed |
| High-Pressure sCO2 Piping | $5-10M | Onshore distribution |
| Heat Exchangers (Titanium) | $8-12M | Server-side and ocean-side |
| sCO2 Pumping Stations | $3-5M | Variable frequency drives |
| Pressure Containment | $2-4M | Safety systems |
| Installation & Engineering | $10-15M | Marine operations |
| Total sCO2 System | $36-61M | $360-610K per MW |
Comparison: Traditional chiller plant for 100 MW: $50-70M
Operating Expenditure (OPEX) Analysis
Power Usage Effectiveness (PUE) by Cooling Method
| Cooling Method | Typical PUE | Cooling Overhead | Annual Cooling Energy (100 MW IT) |
| Air Cooling (Hot Climate) | 1.8-2.0 | 80-100% | 700-876 GWh |
| Air Cooling (Temperate) | 1.4-1.6 | 40-60% | 350-525 GWh |
| Air + Free Cooling (Nordic) | 1.2-1.3 | 20-30% | 175-263 GWh |
| Direct Liquid Cooling | 1.2-1.3 | 20-30% | 175-263 GWh |
| Seawater/SWAC | 1.05-1.15 | 5-15% | 44-131 GWh |
| Ocean sCO2 (NL Concept) | 1.03-1.08 | 3-8% | 26-70 GWh |
Energy Cost Comparison (100 MW, $0.07/kWh)
| Cooling Method | Annual Cooling GWh | Annual Cost | 20-Year Cost |
| Air Cooling (Temperate) | 438 | $30.7M | $613M |
| Air + Free Cooling | 219 | $15.3M | $307M |
| Direct Liquid Cooling | 219 | $15.3M | $307M |
| SWAC | 88 | $6.1M | $123M |
| Ocean sCO2 (NL) | 48 | $3.4M | $67M |
Annual Operating Cost Comparison
| Cost Category | Air Cooling | Ocean sCO2 | Savings |
| Cooling Energy | $30.7M | $3.4M | $27.3M |
| Maintenance | $5.0M | $2.0M | $3.0M |
| Water/Chemicals | $2.0M | $0.1M | $1.9M |
| Refrigerant Replacement | $0.5M | $0.2M | $0.3M |
| Total Annual OPEX | $38.2M | $5.7M | $32.5M |
Total Cost of Ownership (20 Years)
100 MW Hyperscale Data Center - Cooling System Only
| Component | Air Cooling | Ocean sCO2 | Difference |
| Initial CAPEX | $70M | $50M | -$20M |
| 20-Year OPEX | $764M | $114M | -$650M |
| Major Overhauls | $35M | $15M | -$20M |
| Total TCO | $869M | $179M | -$690M |
Ocean sCO2 delivers 79% lower 20-year cooling TCO
Net Present Value Analysis (5% Discount Rate)
| Metric | Air Cooling | Ocean sCO2 | Advantage |
| CAPEX (Year 0) | $70M | $50M | $20M |
| OPEX NPV | $476M | $71M | $405M |
| Overhaul NPV | $22M | $9M | $13M |
| Total NPV | $568M | $130M | $438M |
Heat Transfer Efficiency
| Property | Water | Supercritical CO2 | Advantage |
| Thermal Conductivity | 0.6 W/m-K | 0.08-0.15 W/m-K | Water better |
| Heat Transfer Coefficient | 5,000 W/m2-K | 10,000-15,000 W/m2-K | sCO2 2-3x better |
| Viscosity | 1.0 cP | 0.02-0.1 cP | sCO2 10-50x lower |
| Density | 1,000 kg/m3 | 200-800 kg/m3 | Variable |
| Pumping Power | Baseline | 10x lower | Major advantage |
| System | COP | Meaning |
| Traditional Chiller | 3-5 | 1 kW cooling per 0.2-0.33 kW electricity |
| High-Efficiency Chiller | 6-8 | 1 kW cooling per 0.125-0.17 kW electricity |
| SWAC System | 13-25 | 1 kW cooling per 0.04-0.08 kW electricity |
| Ocean sCO2 (Passive) | 50-150 | 1 kW cooling per 0.007-0.02 kW electricity |
The ocean sCO2 system approaches passive heat rejection - the cold ocean (2-4C) provides the thermal sink, requiring only pumping energy.
NL-Specific Advantages
Ocean Conditions
| Factor | Newfoundland Offshore | Typical SWAC Location |
| Deep Water Temperature | 2-4C year-round | 4-8C |
| Distance to Cold Water | 3-5 km | 1-5 km |
| Water Stability | Excellent (Atlantic) | Variable |
| Infrastructure Access | Oil/gas expertise | Limited |
Enabling Factors
- Abundant Cold Water - Labrador Current provides 2-4C water within 3-5 km of shore
- Subsea Expertise - Hibernia, Terra Nova, Hebron projects built deep-water infrastructure capability
- Low Electricity Cost - $0.03-0.07/kWh makes pumping energy negligible
- Renewable Grid - 97.4% hydroelectric eliminates carbon from any cooling energy used
Risk Factors and Mitigation
| Risk | Probability | Impact | Mitigation |
| Pipe Failure/Leak | Low | Medium | Double-wall pipe, leak detection, CO2 harmless |
| Marine Fouling | Low | Low | Closed-loop eliminates biofouling |
| sCO2 Pressure Loss | Medium | Low | Redundant compressors, emergency venting |
| Permitting Delays | Medium | Medium | Early regulatory engagement |
| Cost Overrun | Medium | Medium | Conservative estimates used |
Technology Readiness
| Component | TRL | Status |
| HDPE Subsea Pipe | 9 | Proven (oil/gas) |
| sCO2 Power Systems | 7-8 | Commercial pilots operating |
| sCO2 Cooling (Industrial) | 6-7 | Research/demo stage |
| sCO2 Data Center Cooling | 4-5 | Concept validation needed |
Comparison with Alternatives
Why Not Conventional SWAC?
| Factor | Conventional SWAC | Ocean sCO2 |
| Corrosion Risk | High (seawater) | None (closed-loop) |
| Biofouling | Ongoing issue | None |
| Pipe Material | Expensive alloys | Standard HDPE |
| Heat Transfer | Good | Excellent |
| Maintenance | Higher | Lower |
Why Not Immersion Cooling?
| Factor | Immersion | Ocean sCO2 |
| Server Compatibility | Limited | Universal |
| Fluid Cost | $50-200K per tank | Minimal (CO2) |
| Heat Rejection | Still needs chillers | Direct to ocean |
| Retrofit Possible | Difficult | Yes |
Implementation Pathway
Phase 1: Proof of Concept ($2-5M)
- Thermal modeling and simulation
- Small-scale sCO2 loop testing
- Ocean temperature surveys
- Regulatory pre-consultation
Phase 2: Pilot System ($15-25M)
- 1-5 MW demonstration at existing facility
- 1-2 km subsea pipe installation
- 12-24 month performance monitoring
Phase 3: Commercial Scale ($50-100M)
- Full 100 MW system design
- 8-10 km ocean loop installation
- Integration with hyperscale facility
Conclusions
- Ocean sCO2 cooling reduces 20-year TCO by 79% compared to air cooling
- Annual savings of $32.5M for a 100 MW facility
- PUE of 1.03-1.08 achievable - industry-leading efficiency
- NL uniquely positioned with cold water, expertise, and renewable power
- Technology risk manageable - all components proven individually
- Payback on incremental CAPEX: <1 year from energy savings alone
References
- Schneider Electric (2020). Liquid vs. Air Cooling CAPEX Analysis
- Electronics Cooling (2023). Supercritical CO2 as Electronics Coolant
- Makai Ocean Engineering. SWAC Technology Overview
- DOE/NETL (2024). Supercritical CO2 Technology Program
- Microsoft (2024). Zero-Water Cooling Systems
- California Energy Commission (2024). Low-Cost Data Center Liquid Cooling
CCNL-CSE-2024-01 v1.0