The sun always shines on TES

Finn Erik Solvang, Ophiolite AS, October, 2026​

Local wasted energy source

From 1 July 2026, newly commissioned data centres in Germany with 300 kW or more of non-redundant connected load are generally required to reuse part of their waste heat, subject to legal exceptions. Many data centres of that size have no obvious local use for the heat, and that gap is what MODERATOR is building hardware for. A 300-kW facility describes a corporate server room, a hospital, a university, or a regional colocation hall. Recovering that heat is worthwhile, but copying the Nordic district heating schemes, Espoo for instance, is not open to most of them, because no network is already there to absorb a constant source. As with any energy recovery system, the issue is matching supply to demand. For that purpose, the MODERATOR project builds an integrated system for thermal recovery, storage and reuse.

Sun absorption for better insulation

Everyone still says data centres get built where it is cold. But CBRE's 2026 analysis lists power availability, grid connection timelines, community approval, fibre, land and tax as the deciding factors, not climate. Atlanta, for instance, hot and humid, is now the largest construction market in North America, and Google's evaporatively cooled sites in Texas and Nevada run at about the same efficiency as its site in Finland. Sunshine in particular does very little to a data centre's cooling bill, because a data hall generates its own heat, so the one absorbed by its envelope is negligible against it. A thermal store is the opposite object: it generates nothing, and every watt it loses, it loses through its skin. So, MODERATOR builds the opposite instinct into the material: a thermal energy storage unit (TES) that recovers waste heat from the data centre, with an outer skin designed to absorb as much sunlight as it can get, so the stored heat lasts longer.

Reverse insulation

Heat loss through an insulated wall depends on two factors: the thermal resistance of the wall and the temperature difference across it. Conventional insulation attacks the first and almost nobody attacks the second. What matters is not the store temperature minus the air temperature, but the store temperature minus the outer surface temperature. Warm the outer surface and the gradient shrinks, whatever the resistance behind it is doing, and heat losses shrink with it. Our topcoat does that on purpose. The outer layer is loaded with coarse magnetite and left to settle as it cures, so gravity grades the particles and the largest and darkest end up at the outer face, giving a solar absorptance of about 0.945. Over it, on 5 mm spacers, sits a pane of recycled photovoltaic glass, leaving a thin sealed air gap that traps infrared against the absorber the way a greenhouse does. Around 87% of the sunlight arriving at the assembly is absorbed. With this design, at 1000 W/m² with 40°C air under a clear sky, that skin settles between 75 and 80°C, depending on the wind. The phase change material inside the store sits at 58°C. For those hours, on those faces exposed to the sun, the gradient has reversed. The TES is not losing heat through them; it is gaining. That does not hold all day, on every face, or in winter; the modelled annual saving is under 10%. But a saving of that order on a seasonal store, won from a surface treatment rather than from more thickness, is not a small result.

Because sunlight reduces heat losses, a unit in a sunny climate reaches the same storage performance on a thinner core than the same unit at Nordic latitude. The design rule that follows is simple: insulation thickness should be adapted to latitude. That matters because here the material is the footprint. Over 85% of the assembly by mass is recycled or recovered, and over 90% can be recycled or reused at end of life, so every millimetre of core you do not need is expanded glass not fired, which means less embodied energy and lower emissions..

Next operations

An instrumented TES insulated with this Ophiolite multilayer, carrying sensors at every layer interface, goes to Greece to sit beside the project's main prototype through six months of real climate exposure. That dataset will let us test the models against measurements.

If you work on data centre heat, on thermal storage, or on insulation that must survive 30 years outdoors, we would like to hear where you think this breaks. MODERATOR runs to the end of 2027.