Cloudberry consists of 12 subprojects


1. Datacenter's interaction with the national energy system
This is a multidisciplinary project where various aspects of data centers interaction with the surrounding energy system, both as a consumer and producer, are studied with regard to possible data center development options. The focus is both on the connection with the energy system, which is analysed, among other things, using the national energy system optimisation model TIMES-Sweden, and on developing an impact assessment of potential data center development opportunities from the perspective of "major socio-technical systems".


2. Multifunctional data centers
This is a collaboration between the architecture- and construction management and building technology  groups at LTU, a data center operator and a construction company, and aims to develop a data-driven demonstrator (MIUC) to evaluate the potential of integrated multifunction data centers and support design and production planning. The tools and processes embedded in the MIUC demonstrator will be developed together with companies, experts and stakeholders to meet their ambitions and make knowledge available to them.


3. Quality and credibility when simulating cooling in data centers
The purpose of the project is to increase energy efficiency in data centers through the development of advanced simulation models and experimental tools. The goal is to develop reliable tools that can be used to optimise and control the cooling of data centers at different levels, from components and subsystems to entire data centers. The produced tools (CFD models) will be used for simulation driven development of cooling methods (free and passive cooling) as well as overall design optimisation. Large scale experiments will be carried out at RISE SICS North Test Facility to validate developed models and further develop data center metering techniques. The research results provide a basis for future projects to develop coordinated models for the cooling system as a whole and energy use in data centers.


4. Transient simulations of thermodynamic connections in data centers
The project aims to develop a simulation tool to capture the dynamics of thermofluidics for the cooling of large-scale distributed IT systems, commonly found in modern data centers. The work is based on an existing simulation tool, RAFSINE, which is based on Boltzmann's method and can run in (near) real-time on graphics processors (GPU processors). With this solution, simulation results can be illustrated graphically, making it possible to study data center layouts interactively and via an optimisation algorithm. A validated, quick, simulation tool will be able to analyse the benefits of different layouts at data centers and heat dissipation methods in case of alternating loads as well as disturbances and errors. In addition, a machine learning component is planned to be introduced to RAFSINE in order to improve the thermal control algorithms for data centers.


5. Evaluation of energy recovery solutions for data centers
A major problem with data centers is that the residual heat generated during the cooling process is rarely used because it often has low temperature and a large volume flow. In this project, possible heat recovery methods will be evaluated in detail where the goal is to visualise which methods and techniques are possible to use as well as whether they are energy efficient enough to become profitable. Another important aspect to study is about and how these affect the operation of data centers, as well as the ability to take advantage of all residual heat. In order to enable this, models will also be developed to predict the amount of residual heat due to the design, operation and geographic location of the data center. Concerning design, it will be studied how different cooling methods such as raised floor, in-row, freezing or liquid cooling affect the potential for heat recovery.


6. Design of distribution networks for energy efficient data center operations
This project aims to review the design of the internal grid within a data center with respect to energy efficiency and reliability. Different distribution network solutions will be studied with a focus on 50 Hz AC and DC power. Solutions with other frequencies and voltage levels will also be studied in detail. New knowledge about opportunities and limitations with energy efficiency of studied options will be developed, as well as existing and expected levels of interference in the data center tension. Studies will also be made of data center performance and data center equipment under realistic interference in voltage (voltage drops, waveform distortion, voltage imbalance and over /under voltage).


7. Data centers impact on the electricity market when the amount of renewable
electricity augment

The project aims to develop knowledge about the consequences for the electricity market in establishing data centers and other energy-intensive industries where the share of renewable electricity generation increases. The goal is to evaluate potential instruments and other financial incentives based on how they can contribute in this context. The pros and cons of a site-dependent network will be studied, as well as the effects of establishing energy-intensive industry and its impact on power grids with increased solar and wind power production. In addition, markets for system services and network services linked to energy intensive industry will be studied, as well as handling balancing power and operational reserves. In conclusion, scenarios with disconnection and management of spare power will be studied as well as a review of which markets or bilateral agreements are needed.


8. SimBerry - Flexible and large-scale simulation of interaction between district heating
and remote cooling networks and data centers

The purpose of the project is to develop methods and tools to dynamically create digital twins of complete district heating and remote cooling networks with recovery of waste heat from data centers. The work is based on previous results where a unique digital twin for a district heating system using an automated model generation method has been created. The goal is to improve flexibility and study how a large-scale digital twin can be made dynamic with respect to functionality and validation facilities. Furthermore, it is studied how the forecasts and thermal storage can be combined to achieve better control and operation. In addition, solutions with machine learning systems for autonomous model update will be studied.


9. Liquid cooling in data centers for direct heat recovery
The project focuses on theoretical and experimental analysis of liquid-cooled IT systems for direct heat recovery. The aim of the subproject is to develop a one-dimensional theoretical flow and heat exchange model that captures both energy and exergy losses when transferred into thermally charged refrigerants. Optimisation of heat path transport routes where the pumping force is minimised (leading to lower so-called PUE2 numbers) will be studied. Another goal is to experimentally validate the theoretical model against a true heat transfer system.


10. Automated maintenance of energy-efficient data centers
Maintenance costs represent a large proportion of total operating costs in large industrial plants such as data centers. Large-scale data centers employ large personnel groups to perform maintenance tasks such as, for example, to replace broken servers and components. The goal of the project is to investigate appropriate methods for replacing people with intelligent and automated server room maintenance systems and measure the impact on data center energy efficiency, as a higher bulk temperature can be allowed if data center is designed for automated maintenance.


11. Resource efficiency in software processes and communication in data clouds
and data centers

The project will study a general software architecture that provides resource-efficient methods for virtualisation of software processes and communication between them in a distributed environment. In particular, so-called process containers are being studied to dynamically distribute computation and storage in order to offer good quality with low energy consumption. The plan is to address energy consumption early in the chain by studying and increasing the efficiency of software. Virtualisation using virtual machines has saved a lot of energy by adding many isolated services to the same physical machine and thus saving energy. In recent years process containers have become an option that is even more lightweight and efficient. A goal for the project is to bring these containers to an additional level of resource efficiency in the context of data centers and distributed devices in society. It involves defining a general architecture for managing and placing software processes. Virtual containers containing processes can be placed to execute in data centers or in devices near the edge of the networks, thereby reducing the amount of communication and resource usage.


12. Resource efficiency in distributed leases and smart contracts
The project will study the distributed distribution area (distributed lists - list of events, distributed in identical copies on many computers) and smart contracts, often including so-called blockchains and hash graphs. These solutions often consume large amounts of energy. The plan is to study resource efficiency in such systems, to define requirements and solutions, to evaluate options, and to define balance between requirements and energy costs. The systems offer that parties (clients) can perform different types of identification and transfers, as well as define mutual agreements through a non-owned third party system, but consist of a number of distributed ownership nodes. These nodes today often consume large amounts of resources (ultimately in the form of energy). The plan for this project is to specifically study resource efficiency in such systems, to define requirements and solutions, as well as to define balances between requirements and energy costs.