The HotFlAd project, funded under grant 03EGB0018A and administered by the Jülich project office, ran from the first quarter of 2019 to the fourth quarter of 2022. The consortium, led by the Hermann‑Rietschel Institute and the Technical University of Berlin, brought together a heterogeneous mix of technical and consulting expertise to develop a novel, climate‑neutral cooling concept for data centers. The core idea is to replace conventional air‑based cooling with liquid‑cooled servers coupled to an adsorption chiller, thereby capturing and reusing the waste heat that is normally released into the environment. The project’s scientific aim was to demonstrate that this integrated system can reduce overall energy consumption, lower the power‑usage effectiveness (PUE) metric, and cut greenhouse‑gas emissions while maintaining or improving IT performance.
Key technical results emerged from a series of computational fluid dynamics (CFD) simulations and prototype tests. The redesigned heat spreader, featuring a serpentine fin geometry and lateral flow paths, achieved a uniform temperature distribution across the server surface, with peak temperatures reduced by up to 15 °C compared with a conventional design. Flow‑velocity maps showed that the new geometry increased the coolant velocity in critical zones by roughly 20 %, enhancing convective heat transfer. When integrated into a full data‑center model, the HotFlAd system lowered the overall cooling energy demand by an estimated 30 % for a 100 kW IT load, translating into a reduction of the site‑wide PUE from 1.8 to 1.5. Energy‑flow analyses for three representative locations—Frankfurt, Madrid, and Stockholm—confirmed that the savings are robust across different ambient conditions. In Frankfurt, the annual cost savings were projected at €250,000, while Madrid and Stockholm showed comparable reductions when adjusted for local electricity tariffs and climate. The adsorption chiller, operating on a water‑based refrigerant cycle, further reduced the carbon intensity of the cooling load by an estimated 40 % relative to conventional vapor‑compression units.
Beyond the technical performance, the project quantified the environmental impact using a greenhouse‑gas‑reduction potential framework. For a nationwide deployment covering all German data centers, the HotFlAd concept could cut CO₂ emissions by up to 1.2 Mt CO₂ eq annually, assuming a 10 % market penetration. These figures were derived from a scalable model that incorporated IT load dynamics, building envelope characteristics, and regional electricity mix data. The study also highlighted that the system’s modularity allows for incremental upgrades, enabling data‑center operators to scale the solution in line with evolving capacity needs.
The collaboration structure was organized around a set of work packages. Work Package 1 focused on defining system requirements, conducting market research, and establishing a business model. A joint requirements workshop in Berlin gathered input from all partners, while surveys and interviews with potential customers informed the market‑potential analysis. The consortium also developed a Sustainable Business Canvas, outlining value propositions such as efficiency gains, scalability, comprehensive service support, and enhanced brand image for climate‑conscious clients. A business‑model workshop held on 19 October 2021 crystallized these concepts and identified key revenue streams and partnership opportunities.
Throughout the project, the consortium maintained close ties with industry stakeholders, including data‑center operators, energy suppliers, and municipal authorities. The project’s outcomes were disseminated through peer‑reviewed publications and conference presentations, ensuring that the HotFlAd concept contributes to the broader dialogue on sustainable data‑center operations. The final report, prepared by the authors, confirms that the HotFlAd system delivers measurable technical benefits while offering a viable pathway for the data‑center sector to achieve significant energy and carbon savings.
