«A Thesis Presented to The Academic Faculty by Steven A. Isaacs In Partial Fulfillment of the Requirements for the Degree Master of Science in the ...»
TWO-PHASE FLOW AND HEAT TRANSFER IN PIN-FIN
The Academic Faculty
Steven A. Isaacs
In Partial Fulfillment
of the Requirements for the Degree
Master of Science in the
School of Mechanical Engineering
Georgia Institute of Technology
COPYRIGHT 2013 BY STEVEN ISAACS
TWO-PHASE FLOW AND HEAT TRANSFER IN PIN-FIN
Dr. Yogendra Joshi, Advisor School of Mechanical Engineering Georgia Institute of Technology Dr. Muhannad Bakir School of Electrical and Computer Engineering Georgia Institute of Technology Dr. Andrei Fedorov School of Mechanical Engineering Georgia Institute of Technology Date Approved: August 5th, 2013
ACKNOWLEDGEMENTSI would like to thank my advisor, Dr. Yogendra Joshi for his guidance and support throughout the graduate process and whose continuous encouragement and mentoring has allowed me to reach my current position in my career. I would also like to thank my committee members, Dr. Muhannad Bakir and Dr. Andrei Fedorov for serving on my committee and their support throughout my research efforts.
I want to thank all of my fellow colleagues in the Microelectronics & Emerging Technologies Thermal Laboratory (METTL) and Consortium for Energy Efficient Thermal Management (CEETHERM) Laboratory. I have been blessed to a part of this lab group and the friendships and intellectual conversations they have provided have been unparalleled. In particular, I want to thank Dr. Yoon Jo Kim for his guidance in construction of the testing platform. I would also like to thank Dr. Muhannad Bakir and Ms. Yue Zhang for fabricating all of the test samples and willingness to work closely with me when I needed assistance. I would also like to thank Dr. Minami Yoda for generously providing access to the high speed camera used throughout this study.
Above all, I would like to thank my family for their support. In particular, I want to thank my mother, Dr. Nelda Isaacs whose passion for academia and unconditional love is the only reason I have made it this far. Lastly, I want to thank my late father, Mr. Mac Isaacs who instilled in me my passion for curiosity and engineering. This work is dedicated to him.
TABLE OF CONTENTSPage ACKNOWLEDGEMENTS iii
LIST OF TABLES vLIST OF FIGURES
5: h vs. q” for varying flowrate for partially heated sample 46 Figure 3.2.
6: h vs. xexit for varying flowrate for partially heated sample 47
Two-phase cooling methods could become the next techniques for high heat removal from high power density electronic packages such as three-dimensionally stacked chips. The small size and unique geometry of such applications makes the existing heat transfer and pressure drop correlations inapplicable. These configurations must be tested experimentally to determine their hydraulic and thermal performance. The focus of the present study is to experimentally determine two-phase performance of surface enhanced micro-gaps.
The beginning of this thesis deals with the introduction of microfluidic cooling methods. In particular microchannel and enhanced micro-gap geometries are considered.
Also, comparison between water and dielectric working fluids is made. A brief overview of flow boiling regime definition is provided, along with relevant flow regime mapping techniques.
Next, the pin fin sample and flow loop testing platform utilized in the present work are discussed. A brief description of the fabrication of the single and multi-heater samples is provided. Unique features of the fabrication and assembly process are described. A detailed description of the setup and operation of the flow loop are discussed. The general experimental procedure provides information on key steps performed for every experiment.
The final section reports the experimental results. A parametric study for each sample is performed by varying heat flux and flowrate. Thermal performance and flow visualization results for both uniformly and partially heat samples are presented and
Modern electronic devices are rapidly becoming more compact and multifunctional. Particularly, with the advent of 3D, stacked architectures, power densities of these devices are continually increasing, much in accordance with Moore’s Law. The need for high heat dissipation cooling methods is crucial and much effort has been exerted in developing these. Microchannel and pin fin enhanced surfaces are commonly utilized in many macroscale heat exchangers. Thanks to the development of micro fabrication techniques, these enhanced features can be easily implemented at the microscale, and are promising options for heat removal from high power density electronic packaging.
Two different categories of working fluids are commonly used for flow boiling based cooling methods, each with its own advantages and disadvantages; water and dielectric fluids . Water has favorable liquid thermal properties, extensive characterization literature and is readily available. Despite its superior heat transfer performance, water has potential disadvantages when used for electronic cooling, if dielectric strength cannot be maintained these include corrosion and possible shorting in the case of leakage. Dielectric fluids, on the other hand, are electrically inert and and can be selected to achieve saturation temperatures closer to maximum allowable chip temperatures, which is not possible with water. Various dielectric fluids such as refrigerants and other novel fluid mixtures continue to be a current research thrust in microelectronic cooling. The following review reveals relevant studies on microchannel
two-phase flow regimes maps is provided.
1.1 Mircochannels Microchannels have been extensively studied within the last 3 decades with focus on determination of heat transfer coefficient, h, as well as critical heat flux (CHF), flow patterns and modeling. Both water and dielectric fluids have been investigated. Water has the advantage of a relatively high thermal conductivity; however, it can be detrimental to an electronic system if leaks develop. Dielectric fluids are non-conductive and are often a more realistic, direct-contact method for cooling from an application standpoint.
In 1981, the pioneering study of single-phase cooling with microchannels was conducted by Tuckerman and Pease . This work demonstrated the low thermal resistance that can be achieved with liquid cooling through microchannels using water.
However, single-phase flows are associated with large temperature gradients along the channel length and, accordingly, larger flowrates. Table 1.1 shows a comparison of single and two-phase flow studies in microchannels. A major consequence under single-phase conditions is a larger associated flowrate relative to two-phase conditions for identical heat removal. In single-phase conditions heat is transfered via sensible heat resulting in a large temperature rise along the microchannel. Two-phase conditions rely on heat transfer via latent heat, in which fluid temperature remains nearly uniform during the boiling process. Therefore, for an identical heat flux a microchannel operating under two-phase conditions requires a lower flowrate and results in a lower surface temperature relative to a microchannel operating under single-phase conditions.
Bowers and Mudawar investigated two-phase flow through minichannel and microchannel heat sinks, reporting a critical heat flux (CHF) above 200 W/cm2 . A comprehensive review of flow boiling in microchannels can also be found in literature , , . Numerous studies involving the modeling of heat transfer and bubble growth have also been performed. For example, Thome et al developed a model to predict local h during slug flow , . Mukherjee et al developed a numerical model to predict bubble growth . A major drawback to two-phase cooling in microchannel heat sinks is flow instability. Qu and Mudawar reported flow instabilities due to pressure drop oscillations that resulted in pre-mature CHF . Numerous papers have been published concerning the suppression and stabilization of this phenomena , , , . A comprehensive evaluation of microchannel cooling methods can be found in literature .
Since flow through microchannel arrays comes with inherent complexities like high temperature gradients and flow instabilities, micro-gaps utilizing augmentation features such as micro pin fins have surfaced as a promising alternative.
To date, a limited amount of literature on the topic of heat transfer and flow over micro pin fin arrays within micro-gaps exists. However, this is a quickly growing area of research. Evaluation of micro pin fin arrays shows a potential advantage over microchannel configurations , , . According to Peles et al, at a similar pressure drop and heat flux, a micro pin fin heat sink provides a minimum total thermal resistance of 0.0389 K/W while a microchannel heat sink provides a minimum of 0.0900 K/W . Experimental values of h near 55 kW/m2K were recorded for single-phase deionized water . Nusselt number correlations for large scale pin fin geometries were compared and observed to over-predict the experiemental data by a factor as high as 2 for low Reynolds number flows (~100). Endwall effects between the pin fins and adjacent walls of the channel imposed boundary layers within the array. With a microscale channel and fin heights, these boundary layers were attributed to the reduced Nusselt numbers at low Reynolds numbers. Suppression of flow separation was also identified as negatively impacting h in smaller devices.
Qu demonstrated decreased thermal resistances using pin fin arrays with strong dependence on liquid flowrate but highlighted the coupled higher pressure drops compared to microchannels . Another comprehensive experimental study identified a lower thermal resistance for a given flowrate using staggered pin fin enhancements compared with other geometries such as inline pin fin, parallel plates, and microchannels . Using a device that was based on vertically integrated chip stacks and that contained electrical interconnect-compatible pin fins, heat fluxes 200 W/cm2 were dissipated at a maximum junction temperature of 80oC considering double-sided heating. This study
emphasized that hydraulic performance should also be stressed when evaluating effective types of pin fin geometries. Specifically, a tradeoff exists between thermal and hydraulic performances . Though pin fin geometries that promote flow separation and mixing result in lower thermal resistances, a high pressure drop is encountered, negatively impacting pumping power. Accordingly, the staggered pin fin orientation has moderate thermal and hydraulic performance and may be a more realistic geometry in terms of application.
One of the earliest studies on using a dielectric fluid flow over micro pin fin arrays looked at single-phase and flow boiling inception . Using a 1,800μm x 10,000μm array of 100μm diameter staggered fins and gap height of 243μm, indicated Nusselt Number values greater than 20 using refrigerant R-123 as a working fluid. Qu et al demonstrated h as high as 180 kW/m2K with 200μm x 200μm staggered square pins with a height of 670μm . Krishnamurthy et al reported a local h as high as 75 kW/m2K using a bank of staggered, circular pin fins with a diameter of 100μm and height of 250μm. It should also be noted that, instead of considering these flow passage enhancements separately, studies have also delved into a combination of microchannel and micro pin fin enhancement . One primary application of pin fin structures is in 3D chips stacks in which pin fins also serve as through silicon vias (TSV) for electrical connections between individual tiers .
1.3 Flow Boiling Regimes In conjunction with thermal and hydrodynamic studies, flow morphology is a key factor in completely defining the particular heat transfer mechanisms that occur for given
clear description of how heat is transferred between the array and fluid. For macroscale sizes, there are generally six separate flow patterns identified with flow boiling through a horizontally oriented heated tube and are shown in Figure 1.3.1 . These flow patterns are also used to describe two-phase flow through channels of various geometry including pin-fin arrays.
At low vapor quality, small, discrete bubbles develop at bubble departure sites, detach from the heated surface and are entrained within the liquid phase. These vapor bubbles are small relative to the size of the tube. This is termed bubbly flow. As vapor quality increases, these bubbles begin to coalesce and these larger vapor bubbles are closer in size relative to the channel and tend to travel along the top of the tube. While these vapor bubbles travel through the tube they are separated by liquid slugs. A liquid film separates the vapor bubbles from the tube wall. This is termed plug flow. A flow regime termed stratified flow is observed for low liquid and vapor velocities in which liquid resides along the bottom of the tube and vapor along the top. The liquid-vapor interface is smooth. As vapor quality increases, the liquid-vapor interface transitions from a flat,