Thermal simulations in support of multi-scale co-design of energy efficient information technology systems

Abstract

Purpose

Information technology (IT) systems are already ubiquitous, and their future growth is expected to drive the global economy for the next several decades. However, energy consumption by these systems is growing rapidly, and their sustained growth requires curbing the energy consumption, and the associated heat removal requirements. Currently, 20-50 percent of the incoming electrical power is used to meet the cooling demands of IT facilities. Careful co-optimization of electrical power and thermal management is essential for reducing energy consumption requirements of IT equipment. Such modeling based co-optimization is complicated by the presence of several decades of spatial and temporal scales. The purpose of this paper is to review recent approaches for handling these challenges.

Design/methodology/approach

In this paper, the authors illustrate the challenges and possible modeling approaches by considering three examples. The multi-scale modeling of chip level transient heating using a combination of Progressive Zoom-in, and proper orthogonal decomposition (POD) is an effective approach for chip level electrical/thermal co-design for mitigation of reliability concerns, such as Joule heating driven electromigration. In the second example, the authors will illustrate the optimal microfluidic thermal management of hot spots, and large background heat fluxes associated with future high-performance microprocessors. In the third example, data center facility level energy usage reduction through a transient measurements based POD modeling framework will be illustrated.

Findings

Through modeling based electrical/thermal co-design, dramatic savings in energy usage for cooling are possible.

Originality/value

The multi-scale nature of the thermal modeling of IT systems is an important challenge. This paper reviews some of the approaches employed to meet this challenge.

Citation

Joshi, Y., Barabadi, B., Ghosh, R., Wan, Z., Xiao, H., Yalamanchili, S. and Kumar, S. (2015), "Thermal simulations in support of multi-scale co-design of energy efficient information technology systems", International Journal of Numerical Methods for Heat & Fluid Flow, Vol. 25 No. 6, pp. 1385-1403. https://doi.org/10.1108/HFF-08-2014-0242