- Format
- Pages
- Chapters
NUMERICAL INSIGHTS OF FLUID-THERMAL CHARACTERISTICS IN A HYBRID MICROJET LIQUID COOLED HEAT SINK
In the current era of immense technological advancement, most of the systems are automated and controlled by electronic components, such as chips and circuit boards, which facilitate better sensing and control in engineering applications varying from data centers to gasturbine engines. In addition, with the fast-paced development track of electronics, the power consumption per unit area has relatively gone up. To complement better operation and longevity of such components, there is a vital need of thermal management methods, having active liquid cooling as its key contributor. The assessment of cooling methods is catered by using computational and experimental methods, where the fundamental parameters are pumping power and thermal resistance. The current study reports on a computational investigation using 3-D CFD simulations aimed at testing nine different hybrid cold plate designs. Constructal theory-based flow channel geometries were used to improve on the hydrodynamic performance in terms of low pressure drop and uniform flow distribution in a fractal branches manifold. High-power dissipation was targeted at a rate of 150 W, emulating a state-of-the-art computer chip. A conventional two-cavity micro jet cold plate was presented as the benchmark for comparison with our hybrid constructal-based micro jet cold plates. The hybrid cold plate configurations used a combination of microchannels and micro jets, have a lower pressure drop than the benchmark configuration by a minimum of 40%. Furthermore, a minimum reduction of 7.5% was noted in the thermal resistance. Seven, out of nine variants have shown promising cooling performance. The implementation of highly efficient flow distribution channels, in combination with the effective cooling mechanism of jet impingement, allowed us to develop a promising alternative for thermal management of high-power electronics
for a wide range of applications.
Chapter 1
1.1 Introduction to thermal management
Thermal management is an integral part of several thermodynamic systems in the modern world. Common examples of thermal management practices range from re-entry vehicles, ICEngines, Gas-Turbine engines, transformers, and data centers to interconnect levels of thermal management. Past years have seen significant contributions by the scientific community in the form of experimental and numerical studies focusing on efficient thermal management of such systems. However, in a scenario where technology has seen a fast-paced growth, there is a substantial need of novel thermal management techniques, which cater for high heat flux management efficiently.
In modern times of highly automated systems, it is hard to imagine a world without electronics, such as PCB boards, computer chips, server racks, LED screens, etc., with varied thermal environments of operation. Common applications of such systems range from ground to underwater applications and airborne environments, such as those of data centers, computer chips, LED arrays, submarines, airplanes, satellites, and space vehicles. It must be noted that such applications have a multifaceted environment for electronics operation and hence thermal management plays a vital role in optimum operation. For instance- the variation of temperature for airplanes is as much as 60-80 ºC within a time span of 15 minutes with an altitude of about 30,000
Choosing the cooling methodology for any application is mainly directed by the amount of heat dissipated by the system and maximum allowable component temperature, while being in optimal range of operation. Typically, cooling is classified as active cooling and passive cooling depending on the mode of heat transfer occurring in the application. Some modern-day applications along with their cooling mechanisms are explained in the following section for better understanding.
1.1.1Applications and their cooling mechanisms
–Avionics cooling– The avionics bay in commercial airplanes is cooled by using pressurized air, which takes away excessive heat by forced convection (active cooling). The air is treated for moisture removal before being sent to avionics bay.
–Space vehicles electronics cooling– The electronics in a space vehicle are cooled by using a liquid coolant (ammonia) in a closed loop active cooling system with a space radiator at 0 K.
–Ship and submarine electronics cooling-Thermal management under water is facilitated by water-cooled heat exchangers wherein air is cooled in a closed or open loop air to water heat exchanger. The airflow in this case is established using a fan and the mode of cooling is active cooling
–Heat pipes and heat spreaders– Heat pipes and heat spreaders are two effective methods used for passive cooling of electronics. Heat spreaders are essentially favorable for
cooling components such as RAM (Random Access Memory) using natural convection, while heat pipes contribute in cooling computer chips using two-phase heat transfer.
–Data center cooling– Data center cooling is achieved by the mode of forced convection wherein the coolant is conditioned air, which is strategically supplied to the facility containing the units to be cooled.
1.2 Motivation for the current project
The previous section was essentially aimed at highlighting the spectrum of engineering applications demanding efficient thermal management solutions. The current section is dedicated at sharpening the focus towards the motivation of the current problem, which is intended at enhancing thermal management in high heat flux microelectronics such as CPUs, GPUs and LED
arrays.
Figure 1-1 Heat generation variation at different length scales for a data center [5] .
In the year 2012, Joshi described to the scientific community the variation of heat generation levels at different length scales in a data center as noted in Figure 1-1[5] . It must be understood that heat generation initiates at the interconnect level and progresses over subsequent length scales presenting the challenges of thermal management in such information technology systems. However, the interconnect level thermal management is comprehended using mainstream nanoscale research as pointed out by Pop [23] . In the light of emerging trends, it is expected that approximately 73 billion kWh/year will be expended by data centers in the United States by the end of the decade [24] . Additionally, 60% of the energy will be used in thermal management of data centers [25] . Hence, thermal management of such budding technologies becomes largely imperative. The motivation for the current study has been attained from operation of highfrequency electronic devices that undergo irreversible transformation of thermal energy that must be dissipated to complement optimum performance of the devices. The chip, in this case takes an important position due to its fast–paced development track over the past years. The additive techniques used in the microfabrication of such electronic devices impose two main challenges to the excess heat removal: 1) several interfaces confined in a micro- to nanoscale space are formed during the microfabrication process, these interfaces are sources of thermal resistance; and 2) the thermal conductivity of the nanoscale components is reduced due to size-effects. Nonetheless, the current study is focused on chip level.
A chip is a combination of various transistors and memory caches leading to formation of independent processing units or cores and hence, it becomes more important to comprehend and develop thermal management methodologies at the chip level. Conventional methods involve usage of active liquid and air-cooling systems, which can be seen in Figure 1-2 and Figure 1-3, respectively, where better thermal performance is noted for liquid cooling systems.
Figure 1-2 Active-liquid cooling [26] . Figure 1-3 Air cooling [27] .
The core of the liquid cooling system is the cold-plate (also termed as water block) which has microstructures in the form of micro channels or fins etched on its surface for effective heat mitigation using the liquid (coolant) which is generally deionized water or any other suitable coolant.
1.3 Objectives of the current study
The current study presents a parametric optimization analysis made on a novel hybrid configuration of cold plates using 3-D CFD simulations. The objective of the present study is based on investigations about the feasibility of the hybrid configuration, which utilizes a combination of micro jet and micro channels available in nine variants. The methodology involves using constructal theory to develop three major categories of channels namely, 16, 32 and 64, which are further bifurcated into three variants each on the basis of inlet diameter. These designs are tested under uniform flow and heat flux conditions aiming to simulate the cooling process of a CPU chip. For reference, a schematic of a commonly used commercial micro channel cold plate is illustrated in Figure 1-4. It has been found that the proposed novel hybrid configuration delivers better fluidthermal characteristics than the benchmark. Additionally, schematics of benchmark configuration and one of the 16 channels variants of hybrid configurations have been presented for better understanding in figures 1-6 and 1-7 respectively.
Figure 1-4 Commercially Figure 1-5 Novel configuration
available cold plate. 2-D schematic.
Figure 1-6 Benchmark configuration. Figure 1-7 16 Channels configuration.
Although, there have been significant contributions to scientific community highlighting the improvement of existing configurations aimed at achieving optimized configurations using micro channel, pin-fin and micro jet cold plates both experimentally and numerically. The objectives of the current study are laid keeping in mind the shortcomings of the existing contributions to the scientific community with an aim to achieve the following:
- Novel design- The design used in the current study incorporates a ‘hybrid’ methodology wherein a combination of micro jets and microchannel is used. The corresponding advantages of the design are presented in the subsequent chapters
- Higher heat flux- With the fast-paced advancement encountered in the chip technology, the thermal design power has seen increments, which demand catering of high heat flux values. The current design is sufficient to cater for thermal design power of 150W.
- Uniformity of flow- As mentioned earlier, uniformity of flow plays a vital role in complementing uniform cooling and avoiding thermal stresses. Flow uniformity was achieved in the current study as one of the main objectives.
- Optimization over existing cases- The current study was undertaken with an aim to provide a thermal management solution, which is optimized over the existing methodologies.
1.4 Thesis outline
The thesis is spread over five chapters. Chapter-1 covered a brief overview of thermal management with examples pertaining to aerospace and other aspects of engineering. This was followed by the motivation of the current problem followed by objectives definition of the current study. Chapter-2 provides vital insights into the relevant literature. Chapter-3 is entitled “Solution Methodology” and covers the problem definition, numerical approach, implementation of the numerical solution, and methods for the analysis of the data. Chapter-4 comprehends the presentation of results and discussion. Chapter-5 is meant to discuss important conclusions and presents suggestions for future work.