对电子系统的有效热设计Effective Thermal Design for Electronic Systems

Christian L. Belady, Hewlett-Packard Corporation
Angie Minichiello, Utah State University

绪论Introduction
过去，大家普遍认为热设计就是预测温度并确保边界值符合产品设计要求。但真正有效的热设计却不仅如此。我们相信在产品设计的早期构思阶段，热设计人员不仅重要而且是最必不可少的。他们就是在这一阶段提出预算。另外，热设计人员的又一作用是排除那些仅为了方便或为了今后查究而设定的规范，从而确定真正的设计要求。In the past, the thermal designer's role was seen as one of predicting temperatures and ensuring that reliability limits are met for products. However, the role of an effective thermal designer is much more than that. We believe that thermal designers are both critical and most useful in the early conceptual stages of product design. This is where they provide the most "bang for the buck." In addition, the thermal designer's role is to challenge specifications that are often used merely for convenience or for legacy reasons, and to uncover the true requirements.

无论怎样，系统热设计的最终目的并不只是预测元件的温度，而是要减少因为热的原因而造成的产品设计失败的风险。针对今天使用电源组的电子系统的固有特性，这种风险在折衷设计方法中十分显著，由于无法预料热及可靠性方面的问题，这类方法通常不能按期完成项目设计。In any case, 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. 图1 就是这种说法的很好证明。热设计工程师从一开始就应该是整个设计工作的一部分，而不应该等开发过程到了后期才发挥作用。曲线图主要为了说明在设计初期所做的经验性预测或者一些快速的笔算有时会比在设计后期才开始进行的所有CFD分析工作更有价值。Figure 1 illustrates this concept very well. Rather than wait until the later part of the development cycle (Figure 1a), the designer should be an integral part of the design right from the start (Figure 1b).  The point of these curves is that your intuition or some quick hand calculations early in the design process is sometimes worth more than all of the CFD analysis work at the end of the design process.

图1. 两种方法在热设计风险管理方面的对比图Figure 1. A comparison of approaches used in risk management for thermal design.

因而，热设计就是工程师使用温度和气流预测帮助系统工程设计人员发现潜在风险区域的过程。（图2 就是一个这样的例子）。Thermal design, therefore, is the process by which engineers use temperature and airflow predictions to uncover potential risk areas for the system engineering team (an example of which is shown in Figure 2). 这里关键在于热设计工程师是整个系统工程设计的一部分并且是以适时适当的精确方法开发可行性解决方案的重要环节。The key point here is that the thermal design engineer is an integral part of the systems engineering approach and it is critical that the thermal designer develop feasible solutions in a timely and reasonably accurate manner. 热设计的最终目标是提供符合项目进度及产品要求的优化设计方案。Ultimately, the goal of the thermal design effort is to provide optimal designs that meet or exceed projected schedules and component requirements.

图2. 一个典型的系统工程团队。其中关键是确定设计过程的重要环节并了解基本性能。Figure 2. A typical system engineering team. It is important to identify the key players in the design and to understand the basic features.

在这一过程中有很多工具来辅助热设计工程师，包括热传递相关性，流程网络模型，计算流体力学，以及实验测量技术。高效率综合性的热设计不一定是要选择“最好”的工具，更重要的是以一种能在短时间内获得足够方案的方式来使用这些工具以便系统工程设计团队能够做出适时的设计决策。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 to choose the "best" tool for design, but rather to use the tools in a way that gets an adequate answer quickly so that the systems engineering team can make timely design decisions.

本文所述的是作者采用的一种热设计方法来设计几种企业型服务器。对于每一个案例，其结果都要满足项目开发的进度同时不能改变热解决方案的构成，空气推动或机械封装等。对如何实际应用这一技术感兴趣的读者可以在ITHERM 2002学报中找到详细的设计实例。This article presents the thermal design methodology used by the authors to design several enterprise servers. Relying upon this methodology, we were able to systematically reduce risk throughout the product design cycle for each of the servers. In each case. the result was that the designs met the project schedule while requiring no changes to component thermal solutions, air movers, or mechanical packaging. Readers interested in how this technique was actually applied will find a detailed design example published in the ITHERM 2002 proceedings [1].

热设计方法论Thermal Design Methodology
热设计方法论的框架是在1997年由Biber和Belady首先提出的，后来发展成为我们今天所说的“增强的产品设计流程”。The framework of this thermal design methodology was first introduced in 1997 by Biber & Belady and later evolved into what we now call the  "Enhanced Product Design Cycle"[2]. 根据这种方法，开发周期由三个阶段组成：设计规划，细节设计和硬件测试。在每一阶段，都要使用最适合最便捷的热设计工具，如图3所示。According to this method, the development cycle consists of three distinct phases: Concept Development, Detailed Design, and Hardware Test. During each phase, the most applicable and expedient thermal design tools are used, as proposed in Figure 3. 因此，这种方法采用一种“流动的”设计过程，在第一阶段对使用工具的预测会与下一设计阶段的预测作比较并采用新的工具。用这种方法，每步设计都会验证，而且热设计工程师可以对其早期工作中的假设和经验进行测试。Thus, the methodology promotes a "fluid" process in which the predictions of tools used in the first phase are compared with subsequent predictions as the design enters the next phase and new tools are adopted. In this way, each step is validated and the thermal designers can test their intuition and the assumptions of their earlier work.

图3. 增强产品的开发流程Figure 3. Enhanced product development cycle.

设计规划Concept Development
设计规划阶段是产品开发流程的最初阶段。在这一阶段初期，产品规划还不成熟。这一阶段的特点是由于系统工程团队的代表们不断讨论产品要求改进设计思想使得产品布局和要求不断改变。The Concept Development Phase is the initial stage in the product design cycle. At the start of this phase, the product concept is in its infancy. This phase is characterized by rapidly changing product layouts and requirements as representatives from the systems engineering team meet to discuss requirements and to develop new ideas. 在此，热设计工程师的目的是彻底分析各个环节，同时迅速提出切实的改进设计方案。这一阶段就是要确定一个能够满足或超出所有设计要求的产品设计格局。当这个阶段快要完成时，热设计工程师应完成以下工作：Here, the goal of the thermal designer is to analyze scenarios thoroughly, yet rapidly as well as offer design suggestions for improvement in real time. This phase concludes in a single product layout that meets or exceeds all requirements. At the conclusion of this phase, the thermal designer has:

• 完成初级的可支持单位功耗的封装设计（从板级到机柜）。Completed a preliminary packaging concept (board through cabinet level) that can support the power dissipation of the unit.
• 完成初步的空气动力选择，包括尺寸，数量，位置以及方向。Completed preliminary air-mover selection, including size, number, location, and orientation.
• 确定重要元件的散热器尺寸和安放位置以确保有足够的空间。Sized and placed critical component heat sinks to ensure that adequate volume is available.
• 估计流过产品的所有排气路径的空气温度。Estimated the air temperature rises through all exhaust paths of the product.
• 确定有热风险的区域并提出必要的减小风险的改进设计建议。Identified areas of thermal risk and proposed design improvements necessary to mitigate these risks.

规划发展阶段的热工具：一个最容易被忽视但却十分强大的工具就是老练的热设计工程师敏锐的设计直觉。Concept Development Phase Thermal Tools:  One of the most overlooked yet powerful tools that the seasoned thermal designer has is his or her intuition. 热设计工程师在很短时间内就可能依以往的经验对当前问题作出快速反应，也许这种第一反应不够准确，但这就够好了。In a matter of seconds, the thermal designer can leverage past experiences to provide a "gut" response that may not be accurate but is, in fact, good enough.  这些经验可以对各种设计规划进行实时的调整，为系统工程师提供难以置信的“预算控制”。These "gut" responses allow real-time sorting of many concepts and offer incredible "bang for the buck" for system engineering teams. In order to do this, the thermal designer needs to develop a sense of intuition, which necessitates constant validation by using tools.

Based upon their ease of use, quick solve times, and limited required input data (usually geometry and fluid data only), the common tools used in the concept phase are generalized correlations for heat transfer and fluid flow (i.e., hand calculations), spreadsheets (such as heatsink design optimizers), and Flow Network Modeling (FNM) techniques. FNM tools can be as simple as solving a network of resistances (both thermal and fluid) either with hand calculations and spreadsheets, or by using commercially available tools. Details of FNM techniques can be found in [1].

FNM techniques are easy to master as long as the 2D air flow paths can be clearly defined. In some cases, 2D flow path prediction is not possible and advanced techniques such as CFD or testing may be warranted. Regardless of whether FNM-specific software or spreadsheets are used, solutions are gained within seconds. This is extremely important in this phase of the development cycle.

Detailed Design
Once a single layout has been agreed upon, the Detailed Design phase of work begins in earnest. The thermal designer must now focus on identified areas of thermal risk within the product. Here, thermal analyses become more detailed (and time consuming), while results become more refined. Experimental measurement and construction of mockups of system-critical areas may be required for input into models or to gain information concerning product areas that are difficult to model. The ultimate goal of the thermal designer is to dig deeply into critical areas and to offer design suggestions, which can facilitate an optimal product design that meets or exceeds the project schedule and reliability goals.

Detailed Design Phase Thermal Tools: Tools commonly used in the electronics industry for detailed thermal design are Computational Fluid Dynamics (CFD) and Finite Element Analysis (FEA) solvers. Typically, these types of analyses require longer setup/solve times and more detailed input data. Three dimensional (3D) modeling or simplification of existing 3D Computer Aided Design (CAD) models may be required. User experience is required for the most accurate results. Furthermore, based upon system size and complexity, empirical sub-system flow resistance data may be acquired using flow-bench testing of fans, heat sinks and subsystems. CFD solvers marketed for the electronics industry afford designers the ability to perform heat transfer calculations in addition to fluid flow solutions. FEA analyses may be required to solve for component, interconnect, or board temperatures once global airflow rates or heat transfer coefficients are calculated.

Hardware Test
Once product prototypes are available, the Hardware Test phase begins. The goal of the thermal designer here is to measure critical components and areas of risk experimentally within the product to verify the design. At this point, there should be no surprises. Additionally, measurements are compared to estimates in order to "calibrate" or fine tune earlier models (calibrated models can be used in future studies). These comparisons are used to determine the accuracy of initial predictions and to help the designer develop "thermal intuition".

Hardware Test Phase Thermal Tools: While the most common tool for temperature measurement is  certainly the thermocouple, additional tools, such as thermistors, resistance temperature detectors (RTDs), thermochromic liquid crystals, thermopiles, and infrared imaging techniques, are also available. Air velocity measurement is also valuable during this phase. Classic hot wire anemometers provide the most precise measurements (i.e., speed and direction), but can be fragile, difficult to calibrate, and expensive. Newer "rugged" hot wire multi-channel probes that measure speed and temperature are also available.

Methodology Summary
Figure 4 shows the impact the thermal engineer can make on the product.  The further to the left that issues and solutions are identified, the more positive the impact upon the product and product schedule. At some point, the thermal engineer can have a negative impact (further to the right) by identifying problems late and requesting changes when the design has become relatively fixed.  This is not where the thermal engineer wants to be.  It is our contention the salary/raises afforded thermal engineers will follow this curve as well.

Figure 4. The thermal designer's impact on the product schedule.

Risk Management
Figure 5 illustrates the importance of using the right tool at the right time. The red curve represents the risk level throughout the development cycle if only intuition, simple hand calculations and spreadsheets are used prior to testing the hardware.  The green curve shows the risk level throughout the development cycle if a complete suite of tools is used.  Note, in this case, the methodology approaches the ideal risk curve as a result of using the modeling technique that matches the design's fluidity.

The key to minimizing the risk early is to use the tool that gets a reasonable answer most quickly.  In addition, Figure 5 also shows that in the case of the green curve, the same risk level is reached in about a quarter of the time shown for the red curve. This implies that an engineer using the technique proposed can be four times more productive.

Figure 5. Using the right tools for risk management.

Conclusions
The focus of the proposed methodology is the systematic reduction of product risk through the careful application of available thermal design tools and techniques. When applied to the design of the servers, the methodology has shown to provide adequate and timely results. The key advantages of the proposed methodology are that it:

• Enables a low risk design, which meets project schedules without necessarily having exact temperature/airflow predictions.
• Utilizes an optimum combination of design tools to increase productivity and reduce design time.
• Exhibits no a priori preference for a given design tool, and emphasizes the use of whichever tool makes sense at the time. The key is to get the right data to make the right design decisions.

Christian Belady, P.E.
Hewlett Packard Company
High Performance Systems Laboratory (HPSL)
Angie Minichiello, P.E.
Space Dynamics Laboratory / Utah Sate University

References

1.Minichiello, A. and Belady, C., " Thermal Design Methodology for Electronic Systems", Proceedings of the ITHERM 2002 Conference, May 2002, pp. 696-704.
2.Biber, C. and Belady, C., "Pressure Drop Prediction for Heat Sinks: What Is the Best Method?", Proceedings of InterPACK '97 Conference, Mauna Lani, Hawaii, EEP-VOL-19-2, ASME, 1997, pp. 1829 -1835.