Optimizing component integration in modular data centres

点击翻译按钮获取中文摘要

Data Centres (DCs) are the physical infrastructure behind critical information technology (IT) services thatsupport our digital lives. Thus, they form an indispensable part of 21st century society. The state-of-the-artprocess of establishing these massive infrastructure is a tedious and lengthy manual exercise leading to inherentdesign inefficiencies resulting in unnecessarily high capital and operating costs for end-users. Our industrypartner Cinnos Mission Critical Incorporated has introduced an innovative family of products, called the SmartMission Critical X (SMC-XTM), which are factory-built cabinets to house IT equipment, containing all thesupporting equipment required to run them. However, due to lack of rigorous scientific design methodologiesto understand the performance implications of the overwhelming number of choices of component selection,layout, and inter-connections, SMC-XTM cannot yet compete with the energy efficiency and costs ofcustom-built DCs used for large volume applications. Developing a rigorous methodology is non-trivial due tothe complex inter-component coupling and discrete variations in input conditions that arise from its modularnature. To overcome these challenges, we will develop and validate a novel design optimization methodologyto transform the next generation of SMC-Xs into a globally competitive product line in the worldwide $100B+DC industry. The proposed novel combination of our modelling and optimization approaches is quite general.It can be extended to design other modular systems with strong multiphysics coupling between components, forinstance, in switchgears and computer hardware. This project will develop the technology and tools required totransform this concept into a globally competitive product line catering to the $100B+ DC industry. The resultspromise considerable commercial revenues to Canada, the creation of 2,000+ manufacturing/assembly jobs, aswell as a substantial amount of energy savings.