A System of Systems approach for data centers optimization and integration into smart energy grids

Abstract

This paper addresses the problem of proactive planning and optimizing the operation of a Systems of Systems (SoS) over a time horizon while considering the characteristics of each constituent system and complex interactions among them. We define a mathematical formalism for modeling complex systems composed of a mesh of sub-systems with linear and non-linear behaviors and abstractions like discrete time, atomic systems and interconnection of atomic system. The proposed modeling approach is simple enough to allow fast computations and simulations, and at the same time complex enough to capture the essential features of the real system thus allowing the mapping of proactive optimization problems to Mixed-Integer Optimal Control Problems. The proactive planning uses hierarchical optimization processes that compute predictions and optimization plans at various time granularities, each finer layer plan adjusting and refining the ones with higher granularity. To show case our approach we model a Data Center which is a well-known case of a large scale complex system aiming to plan and optimize its operation to use as much as possible the locally produced renewable energy and optimize its integration in smart grid advanced context. Simulation based results show a reduction of 5% of the carbon footprint and at the same time an increase in profit of more than 14% due to flexible energy shifting.