Summary
- Schneider tested AQUENE graphene nanofluid against a conventional PG25 coolant in a 100kW closed-loop cooling system.
- Laboratory results showed lower approach temperatures at both the CDU and evaporator across multiple loads and supply temperatures.
- A European operating-site field trial is the next step; claimed PUE savings have not yet been demonstrated at facility scale.
Schneider Electric has reported improved laboratory heat-transfer performance from a graphene-enhanced data centre coolant, with the technology now expected to move towards field testing at an operating European site.
The tests were carried out at Schneider’s manufacturing laboratory in Conselve, Italy, on behalf of a hyperscale customer. They compared a conventional 25% propylene-glycol coolant, known as PG25, against PG25 containing AQUENE, a graphene nanofluid developed by Blue Snow Frontier Technologies.
Schneider and Blue Snow Frontier said the modified fluid reduced approach temperatures in both the coolant distribution unit and evaporator by around 30% to 50% relative to the baseline fluid under the tested conditions.
The companies also say the results could eventually support a 10% to 15% improvement in data centre PUE. That figure is a projected facility-level benefit rather than a measured outcome from the laboratory trial, and Schneider identifies field validation as the next step.
Testing across load and temperature
The test programme used three technology-cooling-system supply temperatures — 20°C, 25°C, and 30°C — and three cooling loads: 50kW, 75kW, and 100kW.
At the coolant distribution unit, Schneider reported reductions in approach temperature of between 0.7K and 1.7K depending on load and set point. At the evaporator, the reduction ranged from 0.3K to 1.2K.
Approach temperature is the difference between the two fluid streams at their closest thermal point inside a heat exchanger. A smaller difference generally indicates that heat is being transferred more effectively, potentially allowing the cooling system to operate with warmer water, reduced compressor work, or other efficiency gains.
AQUENE contains suspended, functionalised graphene particles. Graphene has high thermal conductivity, so the intended effect is to increase the coolant’s ability to carry heat between server-side and facility-side cooling circuits.
Nanofluids also create practical questions that do not appear in a short laboratory efficiency result. Data centre cooling loops are expected to operate for years, and operators need evidence on material compatibility, particle stability, filtration, pump wear, corrosion, maintenance, contamination control, and what happens if the fluid is mixed, drained, or exposed to abnormal temperatures.
Efficiency claims move to the facility
The interest in improved coolant performance is being driven by rack density. Direct-to-chip liquid cooling moves heat away from processors more efficiently than air at high densities, but the effectiveness of the entire chain still depends on cold plates, coolant distribution units, heat exchangers, pumps, facility water loops, and heat rejection.
Reducing temperature differences within that chain can create more operating headroom. Warmer supply temperatures can increase the number of hours when compressors can be reduced or avoided, depending on climate and plant design.
That is the basis for the projected PUE benefit, but PUE is a facility-wide ratio rather than a coolant property. A laboratory improvement at a 100kW test loop cannot by itself establish how much total data centre energy use will fall once UPS losses, fans, pumps, chillers, lighting, and other loads are included.
The planned field trial will consequently be much more important than another incremental laboratory result. An operating data centre can expose the fluid to varying IT loads, real maintenance practices, seasonal conditions, and longer runtimes.
Commercial adoption would also depend on whether the efficiency benefit is large enough to justify changing a fluid that operators already understand. Cooling chemistry tends to be treated conservatively because a leak, blocked cold plate, degraded seal, or fouled heat exchanger can threaten expensive computing equipment.
Schneider’s tests indicate that the graphene additive changes thermal performance in a measurable way. The remaining question is whether that advantage remains stable, safe, and economically useful once the fluid leaves a controlled test rig and spends years circulating through production cooling infrastructure.
