Immersed Computing, from data to fluid flows

The majority of energy consumed by the electromechanical and digital infrastructure that makes up the Internet is converted in to heat. Effective removal of this heat has become core business for a number of technology companies. The Dutch based company, Asperitas, has a unique solution that is based on dense immersion of high powered microelectronics into a basin of inexpensive dielectric liquids that relies purely on natural convection in the harvesting of the heat. This solution encompasses a multitude of flows, from the flow of energy (power in and heat out) to the flow of digital information and the complex fluid mechanics of the primary heat transfer mechanism.

This project is designed to help analyse these flows. The systematic analysis of the naturally convected fluid flow of the dielectric in contact with the small scale features of the microelectronics poses a multiscale fluid dynamics problem that will be analysed using very well developed opensource software and validated against prototype system deployments in a configurable test data centre in Birmingham and in the Energy Building Lab at the University of Leeds. The combined energy and computation will require the development of complex network models that include the flow of energy and data into and out of the appropriately located immersed computing infrastructure, namely Edge or Core, which will require linking the thermodynamics to the processing of data in particular in Edge based deployments.