應用風扇噴流衝擊與水冷卻技術於各種熱移除模組之散熱最佳化研究

Thermal Optimization for Various Heat Removal Modules with Fan-Jet Impingement and Water Cooling Techniques

彭振賢(Cheng-Hsien Peng)

指導教授 ： 洪英輝 ； 傅建中

國立清華大學/工學院/動力機械工程學系/博士(2008年)

摘要

In the present study, a series of theoretical and experimental studies on the fluid flow friction and heat transfer characteristics of four types of heat removal device employed in electronics cooling have been conducted. These four types of heat removal device include (1) heat sink module with fan-jet impingement cooling, (2) heat sink/TEC module with fan-jet impingement cooling, (3) cold plate module with water cooling, and (4) cold plate/TEC module with water cooling. Relevant influencing parameters for heat sink and heat sink/TEC modules with fan-jet impingement cooling are the isoflux Grashof number (GrH), ratio of fan/heat sink gap to characteristic length of heat sink base (s/Lb), fan jet Reynolds number (Rein) and input current of TEC (I). The ranges of parameters studied are GrH =3.94x1010~1.54x1011, s/Lb=0~0.38, Rein=4060~7152 and I=2~10A. Relevant influencing parameters for cold plate and cold plate/TEC modules with water cooling are the isoflux Grashof number (GrH), Reynolds number of cold plate (Recp) and input current of TEC (I). The ranges of parameters studied are GrH =2606~5538, Recp=256~3205 and I=2~10A. Their effects on fluid flow friction and heat transfer characteristics have been systematically explored.
For fan-jet impingement cooling, the flow rate impinging onto heat sink decreases with increasing fan/heat sink gap; while the flow rate of by pass increases with increasing fan/heat sink gap. New correlations for evaluating the overall pressure drop, the effective friction factor (fe) and heat transfer performance factor (j) of cold plate module with water cooling are proposed and the ratio of j/fe for the present cold plate with water cooling is kept at a constant, say 0.078. The local effective heat transfer coefficients for heat sink and heat sink/TEC modules with fan-jet impingement cooling are achieved the uniform distributions; and those for cold plate and cold plate/TEC modules with water cooling result in the uniform distributions in the spanwise direction; while, they gradually decrease along the streamwise direction. Similar trends can be found for local effective Nusselt number, local external thermal conductance and local total thermal conductance. The local and average effective heat transfer coefficients for heat sink and heat sink/TEC modules increase with increasing fan velocity or decreasing fan/heat sink gap; while they are independent of convective heat load; and those for cold plate and cold plate/TEC modules increase with increasing mass flow rate, but independent of convective heat load. Similar trends can be found for local and average effective Nusselt number and external thermal conductance. The average total thermal resistances of heat sink and cold plate modules are dominated by the average external thermal resistance. The effects of average external thermal resistance, module heat load and input current of TEC on local and average total thermal resistance of heat sink/TEC and cold plate/TEC modules are significant. As compared with average total thermal resistance of heat sink and cold plate modules, the maximum reduction of average total thermal resistance for heat sink/TEC and cold plate/TEC modules are to 67% and 76%, respectively; the maximum heat dissipation for cold plate/TEC module is 72W, which is higher than 29 W for heat sink/TEC module, in the present parametric ranges observed. In addition, new experimental correlations as well as the equations derived from the RSM models for local and average effective heat transfer coefficients, Nusselt number, external thermal resistance and total thermal resistance in terms of relevant influencing parameters for four types of heat removal device studied have been presented.
Furthermore, an effective semi-empirical method, which combines thermal network models and empirical correlations, has been successfully established and presented for exploring the following three types of thermal performance improvements with the installation of TEC in the heat removal devices: (1) maximum applicable external thermal resistance for given heat load and maximum chip temperature rise, (2) maximum heat load enhancement for given maximum chip temperature rise and external thermal resistance, and (3) maximum chip temperature reduction for given heat load and external thermal resistance. New correlations for evaluating the design criteria for the validity of using TEC on thermal performance improvements have been proposed. Finally, in order to meet several design objectives and multi-constraints simultaneously and effectively, a systematical design optimization method so-called "RSM-SQP" has been successfully presented and applied to the optimal designs for the present four types of heat removal device under multi-constraints.

關鍵字

熱電致冷器 ； 散熱座 ； 冷板

並列摘要

在本論文研究中，針對四種熱移除模組應用於電子冷卻上之流阻與熱傳特性作一系列理論與實驗的探討。這四種熱移除模組包含：(一)在風扇噴流衝擊下之散熱座模組(二)在風扇噴流衝擊下之散熱座結合熱電致冷器模組(三)在水冷卻下之冷板模組(四)在水冷卻下之冷板結合熱電致冷器模組。在風扇噴流衝擊下之散熱座或散熱座結合熱電致冷器模組的熱流性能之相關影響參數分別如下：穩態葛雷雪夫數(GrH)、風扇與散熱座之間隙與散熱座底部特徵長度之比值(s/Lb)、風扇噴流雷諾數(Rein)與熱電致冷器輸入電流(I)，這些參數的探討範圍是GrH=3.94x1010~ 1.54x1011，s/Lb=0~0.38，Rein=4060~7152 與 I=2~10A。在水冷卻下之冷板或冷板結合熱電致冷器模組的熱流性能之相關影響參數分別如下：穩態葛雷雪夫數(GrH)、冷板雷諾數(Recp)與熱電致冷器輸入電流(I)，這些參數的探討範圍是GrH =2606~5538，Recp=256~3205 與 I=2~10A。在研究中亦針對上述之參數在四種熱移除模組之流阻與熱傳特性上的影響作有系統地探討。
在風扇噴流衝擊冷卻部分，空氣衝擊入散熱座的流量會隨著風扇與散熱座的間隙增加而遞減，而由間隙四周流失的流量則會隨著間隙增加而遞增。在水冷卻之冷板模組研究中，本文針對其總壓降、等效摩擦因子(fe)與熱傳效能因子(j)提出了相關之新的經驗公式；在本研究中之j/fe比值則維持在0.078。局部等效熱傳係數在風扇噴流衝擊冷卻下之散熱座或散熱座結合熱電致冷器模組是呈現均勻分佈；而在水冷卻下之冷板或冷板結合熱電致冷器模組局部等效熱傳係數則是在垂直流體流動方向呈現均勻分佈，而在流體流動方向呈現遞減趨勢。在研究中亦發現在局部等效紐塞數、局部外界熱導與局部總熱導也有相同的分佈趨勢。局部和平均等效熱傳係數在風扇噴流衝擊冷卻下之散熱座或散熱座結合熱電致冷器模組會隨著風扇流速增加或是風散與散熱座的間隙減少而增加，但是與對流熱通量無關；而在水冷卻下之冷板或冷板結合熱電致冷器模組局部和平均等效熱傳係數則會隨著流體質量流率增加而增加，但是與對流熱通量無關。在研究中亦發現局部和平均等效紐塞數以及局部和平均外界熱導變化也是有相同的趨勢。對於散熱座與冷板模組之平均總熱阻是受平均外界熱阻所主導；然而對散熱座結合熱電致冷器模組與冷板結合熱電致冷器模組而言，外界平均熱阻、模組總散熱量與熱電致冷器輸入電流都是影響其平均總熱阻之重要參數。相較於散熱座模組之散熱性能，散熱座結合熱電致冷器模組最多可降低平均總熱阻達67%，而其模組最大散熱量可達29瓦；而冷板結合熱電致冷器模組相較於冷板模組，其散熱性能最多可降低平均總熱阻達76%，而模組最大散熱量則達72瓦。另外，本研究亦針對此四種散熱模組之平均等效熱傳係數、平均等效紐塞數、平均外界熱阻與平均總熱阻分別提出了新的實驗經驗公式以及新的反應曲面公式。
本研究中進一步地建立一個結合熱網路理論模型及經驗公式的半經驗方法來評估在下列三種散熱型態下散熱模組中加入熱電致冷器之熱傳性能增益效果：(一)已知模組總散熱量與最大晶片溫升之最大可應用之外界熱阻，(二)已知最大晶片溫升與外界熱阻之最大模組總散熱量，及(三)已知模組總散熱量與外界熱阻之最大晶片溫降。更值得一提的乃是在研究中針對此三種散熱型態分別提出了經驗公式來界定使用熱電致冷器而能有效提升模組熱傳效能之設計範圍。最後，本研究更成功地建立了一套有效且系統化之多設計目標與多限制條件下的最佳化方法(RSM-SQP)，來針對此四種散熱模組做一系列的散熱最佳化設計探討。

並列關鍵字

TEC ； heat sink ； cold plate

參考文獻

[80] Wang, M. P., Chen, H. T., Horng, J. T., Wu, T. Y., Chen, P. L., and Hung, Y. H., 2005, “Thermal Optimal Design for Partially-Confined Compact Heat Sinks,” IPACK2005-73118, Proceedings of InterPACK’05, International Electronic Packaging Technical Conference and Exhibition, San Francisco, USA, July 17-22.

[9] Lin, S. C., Huang, C. L., and Jan, H. B., 1999, “The Numerical and Experimental Analysis of Axial-flow Fans,” The AASRC Aeronautics and Astronautics Conference, Taiwan, pp. 107–117.

[46] Huang, B. J., Chin C. J., and Duang, C. L., 2000, “A Design Method of Thermoelectric Cooler,” International Journal of Refrigeration, 23, pp.208-218.

[26] Hung, Y. H., Chen, Y. Y., and Liu, C. Y., 1999, “Thermal CAD Package for Multilayer Printed Circuit Boards Integrated with Two-Stack Cold Plate,” In Advances in Electronic Packaging 1999, EEP-26-2, pp.2177-2184.

[23] Fang, C. J., Wu, T. Y., Chen, P. L., Chang, S. F., and Hung, Y. H., 2005, "Thermal Computer-Aided Design for Compact Cold Plates," Journal of the Chinese Society of Mechanical Engineers, 26(1), pp.11-18.