Proper Orthogonal Decomposition for Reduced Order Thermal Modeling of Air Cooled Data Centers

Abstract

Computational fluid dynamics/heat transfer (CFD/HT) methods are too time consuming and costly to examine the effect of multiple design variables on the system thermal performance, especially for systems with multiple components and interacting physical phenomena. In this paper, a proper orthogonal decomposition (POD) based reduced order thermal modeling approach is presented for complex convective systems. The basic POD technique is used with energy balance equations, and heat flux and/or surface temperature matching to generate a computationally efficient thermal model in terms of the system design variables. The effectiveness of the presented approach is studied through application to an air-cooled data center cell with a floor area of 23.2 m2 and a total power dissipation of 240 kW, with multiple turbulent convective components and five design variables. The method results in average temperature rise prediction error of 1.24°C (4.9%) for different sets of design variables, while it is ∼150 times faster than CFD/HT simulation. Also, the effects of the number of available algebraic equations and retained POD modes on the accuracy of the obtained thermal field are studied.

Multi-scale Modeling of Electronics Cabinets

“…This brings a limitation to the method whose effect on the results for the data center cell is studied in Samadiani and Joshi. 13 It is concluded that the method can be used as a reliable and rapid predictor to obtain a new temperature field throughout the system, unless the number of components or available thermal information in the form of equations at the component boundaries is very close to or less than the number of dominant modes. This would not typically cause a problem in thermal model reduction of operational data centers with several housed servers if enough numbers of servers have thermal sensors at their inlet/outlet.…”

“…Samadiani and Joshi 13 have presented a simpler POD based method to generate a reduced order thermal modeling of complex systems such as air cooled data centers. In this method, the algebraic equations to be solved for POD coefficients are obtained simply through energy balance equations, heat flux matching, and/or surface temperature matching for all convective components of the complex system.…”

Multi-objective efficiency enhancement using workload spreading in an operational data center

“…Proper orthogonal decomposition (POD) is another observation based method that is capable of predicting the air flow and temperature fields inside a DC much faster than CFD simulation [21][22][23][24][25][26]. Samadiani et al [27] used POD to derive the thermal map of a DC as a function of the CRAC air flow rate.…”

Comparative Study of Energy Performance between Chip and Inlet Temperature-Aware Workload Allocation in Air-Cooled Data Center

“…The artificial neural network is then combined with a genetic algorithm to optimize the cooling air supply. Besides that, proper orthogonal decomposition (POD), as a reduced-order model algorithm, is widely used to derive temperature profiles from observations [15,16]. Samadiani et al developed a POD-based modeling framework to predict the temperature distribution in computer room [17,18].…”