THERMAL DESIGN METHODOLOGY FOR ELECTRONIC SYSTEMS
Angie Minichiello, P.E.
Space Dynamics Laboratory / Utah Sate University 1695 North Research Park Way North Logan, UT 84341
Phone: (435) 797-4712, Fax: (435) 797-4366
Christian Belady, P.E. Hewlett Packard Company
High Performance Systems Laboratory (HPSL) 3000 Waterview Parkway
Richardson, TX 75080 Phone: (972) 497-4049, Fax: (972) 497-4245
This paper presents the thermal design methodology used to design a multi-processor enterprise server, the RP8400. The proposed methodology combines well-known analytical and experimental thermal design tools, including heat transfer correlations, Flow Network Modeling (FNM) and Computational Fluid Dynamics (CFD) techniques, and experimental measurements. The key benefit of this methodology is its emphasis on the use of varied design tools, each applied at its optimal point in the product design cycle. Thus, analysis time is greatly reduced, with acceptable sacrifice to accuracy and detail, during the earliest stages of design when the design concept is fluid, new ideas abound, and speed is paramount. Detailed analyses, providing a greater degree of accuracy, are performed in the latter stages of the development cycle when designs are firm, changes are fewer, and optimization/validation is the goal. In this manner, thermal risk is systematically reduced throughout the product design cycle. This paper begins with an overview of the thermal design methodology. Direct application of the methodology to the design of an enterprise server, the RP8400, is discussed. Numerical modeling and empirical results are presented and compared, followed by a discussion of methods for improving thermal design in future products.
KEY WORDS: FNM, CFD, Thermal Design, Electronics Cooling, System Analysis
The ultimate goal of system thermal design is not the prediction of component temperatures, but rather the reduction of thermally associated risk to the product . This risk, inherent to today’s power-packed electronic systems, is manifested by compromised designs that do not meet projected schedules due to unforeseen thermal and/or reliability issues. Thermal design, therefore, is the process by which engineers use temperature and airflow predictions to uncover potential risk areas and develop feasible solutions as early as possible in the product design cycle. Ultimately, the goal of this effort is to provide optimal designs that meet or exceed projected schedules and component requirements.
Many tools exist to assist thermal design engineers during this process, including heat transfer correlations, Flow Network Modeling (FNM), Computational Fluid Dynamics (CFD), and experimental measurement techniques. The key to efficient and comprehensive thermal design is not, necessarily, the choice of the “best” tool for design, but rather the optimized integration of available tools.
This paper presents the thermal design methodology used to design the RP8400 enterprise server. Possessing a maximum power per area rating of over 800 W/ft2, the RP8400 required careful thermal analyses in order to meet its aggressive project schedule. Relying upon the methodology presented herein, the thermal designer was able to systematically reduce risk throughout the product design cycle, culminating in a design that met the project schedule while requiring no changes to component thermal solutions, air movers, or layout.
A system level thermal design methodology is presented and applied to the design of a multi-processor enterprise server, the RP8400. The focus of the proposed methodology is the systematic reduction of product risk through the careful application of available thermal design tools and techniques.
Applied to the design of the RP8400, the methodology is shown to provide adequate results for the thermal design of large systems possessing complex 3D flow patterns. The key advantages of the proposed methodology are that it:
(1) enables a low risk design which meets project schedules without necessarily exact temperature / airflow predictions
(2) utilizes an optimum combination of design tools in order to increase productivity and reduce design time
(3) exhibits no a priori preference for a given design tool;
emphasizes the use of whichever tool which makes sense at the time
Finally, increased accuracy over reported results can be obtained, as resources permit, through the more frequent application of detailed modeling techniques (CFD, FEA) and sub-system testing.
We thank Mr. Brad Clements, of Hewlett Packard Company, for his support and technical assistance throughout the RP8400 design effort.
标签： 点击： 评论: