scispace - formally typeset
Search or ask a question
Journal ArticleDOI

Electronics Thermal Management in Information and Communications Technologies: Challenges and Future Directions

TL;DR: This paper identifies drivers for progress and immediate and future challenges based on discussions at the 3rd Workshop on Thermal Management in Telecommunication Systems and Data Centers held in Redwood City, CA, USA, on November 4–5, 2015.
Abstract: This paper reviews thermal management challenges encountered in a wide range of electronics cooling applications from large-scale (data center and telecommunication) to small-scale systems (personal, portable/wearable, and automotive). This paper identifies drivers for progress and immediate and future challenges based on discussions at the 3rd Workshop on Thermal Management in Telecommunication Systems and Data Centers held in Redwood City, CA, USA, on November 4–5, 2015. Participants in this workshop represented industry and academia, with backgrounds ranging from data center thermal management and energy efficiency to high-performance computing and liquid cooling, thermal management in wearable and mobile devices, and acoustic noise management. By considering a wide range of electronics cooling applications with different lengths and time scales, this paper identifies both common themes and diverging views in the thermal management community.

Content maybe subject to copyright    Report

Purdue University
Purdue e-Pubs
CTRC Research Publications Cooling Technologies Research Center
2017
Electronics $ermal Management in Information
and Communications Technologies: Challenges
and Future Directions
S. V. Garimella
Purdue University, sureshg@purdue.edu
T. Persoons
J. A. Weibel
Purdue University, jaweibel@purdue.edu
V. Gektin
Huawei
Follow this and additional works at: h=p://docs.lib.purdue.edu/coolingpubs
<is document has been made available through Purdue e-Pubs, a service of the Purdue University Libraries. Please contact epubs@purdue.edu for
additional information.
Garimella, S. V.; Persoons, T.; Weibel, J. A.; and Gektin, V., "Electronics <ermal Management in Information and Communications
Technologies: Challenges and Future Directions" (2017). CTRC Research Publications. Paper 321.
h=p://dx.doi.org/10.1109/TCPMT.2016.2603600

1
Electronics Thermal Management in Information
and Communications Technologies: Challenges and
Future Directions
Suresh V. Garimella, Tim Persoons, Justin A. Weibel, and Vadim Gektin
Abstract This paper reviews thermal management
challenges encountered in a wide range of electronics
cooling applications from large-scale (data center and
telecommunication) to small-scale systems (personal,
portable/wearable and automotive). The paper identifies
drivers for progress, and immediate and future challenges,
based on discussions at the 3
rd
Workshop on Thermal
Management in Telecommunication Systems and Data
Centers held in Redwood City, California, on November 4-
5, 2015. Participants in this workshop represented
industry and academia, with backgrounds ranging from
data center thermal management and energy efficiency to
high-performance computing and liquid cooling, thermal
management in wearable and mobile devices, and acoustic
noise management. By considering a wide range of
electronics cooling applications with different length and
time scales, the paper identifies both common themes and
diverging views in the thermal management community.
Index Terms Information and communication networks;
Energy management; Thermal management of electronics.
I. BACKGROUND
The content of this paper is informed by presentations and
discussions during the 3
rd
Workshop on Thermal Management
in Telecommunication Systems and Data Centers held in
Redwood City, California, November 4-5, 2015. The
workshop drew together representatives from academia
1
and a
range of industries
2
spanning large-scale electronics and
1
Purdue University, Trinity College Dublin, Kyushu University,
University of Houston, and Stanford University
2
Amazon Lab126, CoolIT Systems Inc, EXA Corporation, Fujitsu
Limited (Japan), Hewlett Packard Labs, Huawei Technologies (US,
Sweden and China), Intel Corporation, Qualcomm, Samsung
Electronics, and Toyota Research Institute of North America
telecommunications systems, to small-scale personal
electronics. While the underlying thermal management
challenges are similar in many of these applications, each is
inevitably characterized by different drivers and constraints.
This paper is inspired by discussions and presentations at this
Workshop, and organized as follows: (i) observations related
to the background for ICT thermal management, (ii) the
business and technological drivers for progress, and (iii) a
review of the wide range of thermal management challenges in
large- and small-scale systems, using the workshop
participants’ opinions as a starting point. (iv) Finally, an
overview of the implementation of thermal solutions is
presented, broken down into different technologies and heat
transfer modes.
A number of recent publications have addressed the thermal
management challenges in large-scale electronics systems,
including those by the authors [
1
,
2
] that were based on prior
workshops in this series organized by the Cooling
Technologies Research Center, a National Science Foundation
Industry/University Cooperative Research Center at Purdue
University. Because of the growing energy consumption by
large-scale electronic systems and the data center sector in
particular [1], thermal management challenges in this sector
have received a lot of attention in recent research and review
papers [
3
,
4
]. The current paper acknowledges the significant
challenges and opportunities in large-scale electronic systems;
yet, its aim is to provide a more comprehensive context by
also considering the needs and restrictions in small-scale
electronic systems. The increasing degree of inter-
connectivity, continuing growth in data volumes at the level of
personal electronic devices, and proliferation of embedded
electronics and sensors in transportation systems will
inevitably have repercussions on the energy usage by large-
scale systems providing services such as cloud-based data
storage and computing.
A widely adopted energy efficiency metric encountered in
data center thermal management, and one which therefore

2
appears frequently in this report, is the power utilization
effectiveness (PUE), defined as:








󰆄
󰆈
󰆅
󰆈
󰆆
(1)
in which

,

,
are the power consumption of the ICT
equipment, thermal management and electrical power delivery
systems, respectively.
The amount of annual data generated globally in the last
decade has rapidly increased from 0.1 ZB in 2005, to 1.2 ZB
in 2010, and 8.5 ZB in 2015, with a projection of 40 ZB for
2020 [
5
]. Amazon, Google, and Uber are just a few examples
of market-disrupting businesses that rely on real-time
processing of ‘Big Data’. The handling of vast amounts of
scientific data, such as human DNA sequences, would benefit
enormously from cloud-based data storage and computing [
6
].
In terms of the energy footprint of public cloud computing, a
survey by Uptime Institute has revealed that the average data
center’s power usage effectiveness (PUE) has been only
slightly reduced from 1.89 in 2011 to 1.70 in 2014 [
7
], still
leaving ample room for improvement compared to the ideal
value of unity.
In addition to the traditional air-cooled servers in a data center,
‘hybrid’ thermal management solutions employing a
combination of air cooling and liquid cooling at the server
level are being gradually introduced for high-end computing
facilities. In a hybrid air/liquid cooled server, only the
components with the highest heat loads such as processors are
liquid cooled. More conventional liquid-based heat spreading
techniques (e.g., heat pipes, vapor chambers) are not taken
into consideration here. One key benefit of the hybrid cooling
approach over traditional air cooling is the increased potential
for recuperation of waste heat from the liquid coolant stream.
Different system configurations are being considered to make
use of this waste heat. Some of these involve complex
thermodynamic cycles for maximizing energy efficiency and
computational performance [
8
]. A hybrid liquid cooling
solution (60% by water, 40% by air [8]) is cheaper than a fully
liquid-cooled analog, and yet achieves a PUE as low as 1.3
[
9
]. With a tendency towards density-driven designs, liquid
cooling (even if only used on high-power components such as
CPUs) allows for about three times denser CPU packaging for
the same footprint area, and a similar increase in volumetric
power density (W/m
3
) [9]. Indeed, performance is increasingly
being expressed in terms of volumetric performance
(Gflops/W and Gflops/m
3
) [9]. This will require an even
closer integration of ICT equipment and thermal management
approaches in future designs.
Extreme heat fluxes beyond those encountered in typical data
centers must be dissipated in radar, power electronics, and
high-performance computing (HPC) systems. Evaporative
cooling strategies in direct contact with the semiconductor
device are therefore the focus of the current DARPA
Intrachip/Interchip Enhanced Cooling (ICECool) program
[
10
]. Targeted heat density levels are > 1 kW/cm
2
and > 1
kW/cm
3
with a chip temperature rise below 30C and
maximum temperature difference across the chip footprint
below 10C.
Thermal design for portable electronics requires approaches
that are quite different from the traditional paradigms in large-
scale electronics cooling. Here, the focus is on customer-
centered performance, and thermal targets are driven by
ergonomic considerations such as limits on the skin
temperature. However with increasing performance and
decreasing form factors, this end of the design spectrum also
relies on concurrent interaction between electrical and thermal
engineering teams. Electrothermal co-design emerges as a
common theme for next-generation systems and devices,
whether at large or small scale.
II. DRIVERS FOR PROGRESS
A. Business drivers
Current investment strategies in large-scale ICT facilities seem
to be driven more by operational expenditure (OpEx) than by
capital expenditure (CapEx), as evidenced by the increasing
number of large-scale bare-bones systems relying on free
cooling, designed for resilience rather than redundancy [1].
One might wonder whether relying solely on air cooling could
result in a reluctance to invest in the development of new
technology. Without research and development investment in
higher risk cooling technologies including liquid cooling,
higher CapEx costs may prevent these technologies from
reaching a high enough penetration to achieve sector-wide
savings in OpEx by a reduction in energy use [1]. However,
changes in policy could induce a change in the CapEx balance
between air and liquid cooling. Acoustic noise regulations are
set by standards and thus subject to policy changes. An
increase or decrease in noise emission thresholds would
change the level of detail needed from aeroacoustics
simulations in the development of servers. Along with other
drivers including enhanced reliability and increased energy
efficiency, this may push up the cost of air-cooled systems.
In other applications such as hybrid electric vehicles (HEVs),
the power density of inverters is such that air cooling would
require high fan loads resulting in excessive acoustic noise
levels. This has driven decisions towards liquid-cooled cold
plates in personal vehicles. A similar shift may well arise with
the advent of 3D integrated circuits and denser packages in
HPC and volume servers.
Market drivers for liquid cooling also vary by region.
Computational performance using overclocked CPUs is the
main driver in North America, whereas energy efficiency and
density are the main drivers for Europe and Japan due to the
higher electricity cost.
The market demand for mobile phones with added features,
functionality and performance inevitably increases thermal
susceptibility. To keep up with performance demands, the
average CPU power density per core has been steadily

3
increasing with phone generations. Top market drivers include
price, battery life, functionality, as well as sleekness of the
product which complicates electro-thermal design. Phone
makers have a paradoxical choice concerning sleekness versus
battery life although sleekness is not a top priority based on
customer surveys, manufacturer vie for dominance in the race
for ever-thinner phones.
Comparing maximum power dissipation and the approximate
cost of the thermal solution across different form factors, (i) a
laptop is limited to 15-20 W at a cooling cost of $1.50/W,
while the budget is (ii) 7-10 W and $2.50/W for a tablet and
(iii) 2-3 W and $3.50/W for a mobile phone [9].
Other possible practical business constraints to a further
penetration of energy-saving cooling methods could include
the need to accommodate legacy air-cooled products, and
being able to guarantee customers an established level of
reliability.
B. Technological drivers
1. Computing versus communication costs
A key technological driver is the balance between the cost for
computation versus the cost for communication [1]. This
balance exists on the micro level, e.g., driving the evolution
towards integration of logic and memory chips in a single
package, as well as on the macro level in the balance between
(on-device) local computing versus (off-device) cloud
computing in data centers.
It is worth asking if there may be a shift back from cloud
computing toward local computing if cheap, high-density
options for local computing became available (such as IBM’s
concept for a sugar cube-size supercomputer [
11
]). Or would
this lead to data centers simply adopting and multiplying this
enhanced technology on a larger scale?
It remains impossible to combine all functionality in a single
device; therefore, the distributed heterogeneous model is the
only viable solution. Intel’s Knights Landing is a ‘many
integrated core’ coprocessor chip [
12
,
13
] which is a practical
example of this model.
From a provider’s point of view, it is more economically
viable to carry out as much local computing on the device as
possible, since these energy costs are at the user’s expense,
whereas server energy consumption is at the provider’s
expense.
As transmitted data volumes increase, photonics and
combined electronics/photonics modules for optical
communication will gain in importance. The thermal
management requirements are quite different in this case, with
laser-based photonics requiring very precise temperature
control to prevent laser wavelength drift. However, the same
cooling technologies can be used for both electronics and
photonics packages; for photonics, actively controlled
thermoelectric coolers are considered in addition [
14
,
15
].
Photonics thermal management challenges may become more
extreme for optical power transfer applications [
16
]. The
introduction of more and more sensors in automotive systems
has led to a significant increase in the amount of copper
wiring. Significant weight savings would be achieved if a
sensor could be connected with a single optical fiber for
communication and power delivery. There are currently about
100 sensors in a typical car, but this number will be
significantly higher for autonomous vehicles [
17
,
18
]. Vehicle-
to-network communication is already in place (to exchange
traffic information, for example) but the future will see more
vehicle-to-vehicle communication as well.
2. Proliferation of different computing systems platforms
At present we have a good grasp of the thermal behavior of
components; the computing revolution continues, however,
with new and as yet unimagined applications in the Internet
of Things (IoT) [
19
]. As new applications emerge, the focus
has shifted to systems rather than components. Computing will
evolve along three vectors: (i) small, flexible, light-weight
interconnected devices; (ii) increased computing performance
at manageable cooling costs in data centers and other server
markets; and (iii) increased proliferation of embedded
computing (e.g., in automotive, healthcare, buildings, space,
and other commercial/consumer markets).
The number of devices has increased from 2 billion in 2006 to
15 billion in 2015, and is projected to approach 20-50 billion
by 2020 [9,
20
,
21
]. This evolution will see a rise in background
computing in data centers. Today, the data center business
includes enterprise ICT, cloud service providers,
telecommunication service providers and scientific computing.
From 2014 to 2018, a 15% growth in data center business is
expected [9]. The main growth drivers are (i) the cloud market
offering low-cost quick access, (ii) high performance
computing (HPC), and (iii) Big Data manipulation.
From the 1990s to 2015, devices have evolved from clunky to
sleek systems with more functionality, becoming slimmer and
more portable. The next step seems to be even thinner, flexible
devices which could be integrated in new ways, e.g., as part of
clothing in the form of wearables. This broad spectrum of
devices, packages and form factors will see a growing set of
communication protocols. If different devices should be able
to talk to one another, they will have to evolve an ability to
support multiple protocols.
Across the spectrum from portable devices to data centers, a
key focus is energy. Intel’s Shekhar Borkar anticipates an
exascale data center with a power consumption below 20 MW
to be realizable by 2020 [
22
]. This achievement would require
a hypothetical processor with 4,096 cores on a die capable of
16 teraflops at double precision for below 100 W, or 200
Gflops/W. This exascale machine could be realized by
combining 62,500 of these processors with four threads per
core.

4
3. Thermal management for mobile phones
Experts on mobile phone thermal design see two clear trends:
(i) towards thinner phones, requiring thinner electronics
packages and boards, with package thickness reducing from
1.4 mm to 1 mm, and further to 0.5 mm in the near future; and
(ii) customer demand for longer battery life that results in an
increase in battery size [9]. However, because the printed
circuit board (PCB) size has not changed, these trends lead to
a consideration of different internal layouts. The PCB takes up
30-40% of the phone by area. Typical layouts have the battery
below and PCB on top, or a C-shaped PCB with an elongated
battery alongside. The board has very dense routing and this
leads to a risk of hot spots on the outer skin surface of the
phone within the limited package size.
Portable consumer electronics such as mobile phones have
seen significant evolution in performance and to a lesser
extent also in battery capacity. Between 2010 and 2012, the
processor performance for Samsung Galaxy models has
increased 5.9 times while the battery capacity increased 1.4
times [
23
]. The trends are similar for the Apple iPhone. So
while performance and heat dissipation are sharply increasing,
battery capacity is not following at the same rate.
The trend towards smaller form factors combined with higher
performance means that hot spots become more prevalent and
problematic to resolve. It is becoming more important to
balance computing performance and power consumption. Key
factors for innovative, active thermal solutions include heat
radiating capacity, noiseless and vibration-free operation,
scalability to smaller form factors, reliability, and low cost.
For wearables, low-power operation is crucial both from the
perspective of battery life and ergonomic constraints related to
skin temperature.
4. How to extend the life of air cooling?
At a previous Workshop in this series in 2012, the following
list of required developments was proposed to extend the life
of air cooling in data centers [1]:
1. At the system level: More efficient air movers
(coefficient of performance, COP > 20) and a
reduction in fan noise to below 60 dB(A).
2. At the board level: Optimized heat sinks with low
pressure drop and thermal resistance,

(e.g., <
0.1C/W for a 140 CFM 1U volume server) and
alternative air movers to decouple pumping power
and heat transfer, such as piezo fans, synthetic jets,
and electro-aerodynamics.
3. At the component level: Limit the component power
dissipation and reduce

for component and
thermal interface material (TIM).
Today there still seems to be general agreement on these
targets, with particular emphasis on the importance of
component-level developments. The system- and board-level
developments are determined by economic drivers but the
same board-level targets were deemed valid at the 2015
Workshop as well [9].
Regarding acoustic noise in data centers, there are two
limitations to be considered: (i) annoyance and ergonomics,
and (ii) regulatory safety limits for maintenance workers.
While there is debate over the necessity of the former in data
center environments, the latter will always remain because
data centers cannot afford to shut down to carry out
maintenance in quiet conditions. In residential areas, the
annoyance limitation may yet prove important.
A significant contribution to the noise problem in air cooling
comes from excessive fan backpressure resulting from poorly
designed grilles and backplane connectors. Even in a notebook
computer, the pressure drop is split approximately equally
between the heat sink and the grilles and ducting.
Improvements are possible by simply arranging connectors
differently. Part of the solution would be to better educate
other design teams on thermal-fluid engineering and
aeroacoustics. Fans are already being customized for certain
applications, with increased dialogue between ICT systems
manufacturers and fan suppliers. This tailored design approach
should include the placement of the fan and all components
making up the air path.
A sometimes overlooked contributor to fan load is the need for
air filtration, especially in uncontrolled building environments
located in emerging markets with often high levels of airborne
pollutants. The pressure drop across the filters may cause the
fans to operate closer to stall conditions, thereby further
increasing power consumption and acoustic noise levels
[
24
,
25
,
26
].
For certain high-density board layouts, air cooling may
become so challenging that liquid-cooled cold plates are the
only viable option. The three main drivers for liquid cooling
are performance, efficiency and density. Most practical
implementations would not be 100% liquid-cooled, but instead
involve a hybrid combination of liquid cooling for high-power
components and air cooling for low-power components. This
may shift the backpressure balance in favor of either fans or
alternative air movers, although many of the latter (e.g., piezo
fans and synthetic jets) can overcome only a small pumping
head and require an integrated design approach to provide
sufficient steady-state cooling performance [
27
,
28
].
In mobile devices, widespread introduction of forced air
cooling is unlikely. However in an IoT context, there could be
a need to plug in a mobile device into docking stations which
could provide active cooling to unlock increased performance
in the device. A large contact area with low thermal resistance
would be required to accomplish this or some other form of
thermal connection. At least for now, handheld devices do not
seem to have reached a plateau where active cooling has
become essential. Power-hungry functionality such as video
recording is already commonplace, and other limits such as
battery life are often hit before thermal limits.

Citations
More filters
Journal ArticleDOI
TL;DR: In this paper, physics-informed neural networks (PINNs) have been applied to various prototype heat transfer problems, targeting in particular realistic conditions not readily tackled with traditional computational methods.
Abstract: Physics-informed neural networks (PINNs) have gained popularity across different engineering fields due to their effectiveness in solving realistic problems with noisy data and often partially missing physics. In PINNs, automatic differentiation is leveraged to evaluate differential operators without discretization errors, and a multitask learning problem is defined in order to simultaneously fit observed data while respecting the underlying governing laws of physics. Here, we present applications of PINNs to various prototype heat transfer problems, targeting in particular realistic conditions not readily tackled with traditional computational methods. To this end, we first consider forced and mixed convection with unknown thermal boundary conditions on the heated surfaces and aim to obtain the temperature and velocity fields everywhere in the domain, including the boundaries, given some sparse temperature measurements. We also consider the prototype Stefan problem for two-phase flow, aiming to infer the moving interface, the velocity and temperature fields everywhere as well as the different conductivities of a solid and a liquid phase, given a few temperature measurements inside the domain. Finally, we present some realistic industrial applications related to power electronics to highlight the practicality of PINNs as well as the effective use of neural networks in solving general heat transfer problems of industrial complexity. Taken together, the results presented herein demonstrate that PINNs not only can solve ill-posed problems, which are beyond the reach of traditional computational methods, but they can also bridge the gap between computational and experimental heat transfer.

261 citations

Journal ArticleDOI
TL;DR: In this paper, the thermal efficacy of half-sinusoidal non-uniform heating at different spatial frequencies for a porous natural convection system using Cu-Al2O3/water hybrid nanofluid and magnetic field was examined.
Abstract: The present work aims to examine the thermal efficacy of half-sinusoidal nonuniform heating at different spatial frequencies for a porous natural convection system using Cu–Al2O3/water hybrid nanofluid and magnetic field. The system is presented utilizing a classical square enclosure heated nonuniformly at the bottom wall, and the sidewalls are allowed to exchange heat with the surroundings. The Brinkman–Forchheimer–Darcy model is adopted catering other additional terms for buoyant force and magnetic field. The governing equations are transformed into nondimensional forms and then solved numerically using a finite volume-based computing code. The importance and fundamental flow physics are explored in terms of the pertinent parameters such as the amplitude (I) and spatial frequency (f) of half-sinusoidal heating, Darcy–Rayleigh number (Ram), volume fraction of hybrid nanoparticles ( $$ \phi $$ ), and Hartmann number (Ha). The flow structure and heat transfer characteristics are analyzed and presented utilizing heatlines, streamlines and isotherms and average Nusselt number. The results show that the use of half-sinusoidal nonuniform heating along with hybrid nanofluid can be a viable method for enhancement and control of the overall thermal performance. The study indicates that half-sinusoidal heating could be a promising technique for better heat transfer even in the presence of flow dampening effects like porous media and magnetic fields.

83 citations

Journal ArticleDOI
TL;DR: In this article, the authors present a brief review of thermal enhancement using synthetic jet along with various parameters that influence its flow-field and cooling performance, and a list of potential gaps and challenges are presented for laying down the foundation for future research.

71 citations

Journal ArticleDOI
TL;DR: In this paper, the authors demonstrate an approach for augmenting heat transfer through porous media subjected to nonuniform heating during the magnetohydrodynamic flow of a hybrid nanofluid of Cu-Al2O3/water.
Abstract: The intent of this study is to demonstrate an approach for augmenting heat transfer through porous media subjected to nonuniform heating during the magnetohydrodynamic flow of a hybrid nanofluid of Cu–Al2O3/water. The efficacy of such a heating technique is examined utilizing a classical flow geometry consisting of a square cavity. The heating is made at the bottom following a half-sinusoidal function of different frequencies, along with the presence of a uniform magnetic field. The thermal conditions of the cavity, particularly at the bottom wall, drive thermo-hydrodynamics and associated heat transfer. Furthermore, the addition of different types of nanoparticles to the base liquid in order to boost the thermal performance of conventional fluids and mono-nanofluids is a current technique. The coupled nonlinear governing equations are solved numerically in dimensionless forms adapting the finite volume approach, the Brinkman–Forchheimer–Darcy model, local thermal equilibrium and single-phase model. The study is conducted for wide ranges of parametric impacts to analyze global heat transfer performance. The results of this study reveal that the multi-frequency spatial heating during hybrid nanofluid flow can be utilized as a powerful means to improve the thermal performance of a system operating under different ranges of parameters, even with the presence of porous media and magnetic fields. In addition to different heating frequencies, the variations in amplitude (I) and superposed uniform temperature ( $$\theta_{\text{os}}$$ ) to half-sinusoidal heating are also examined thoughtfully in the analysis for different concentrations of Cu–Al2O3 nanoparticles. Compared to the base liquid, the hybrid nanofluid can contribute toward higher heat transfer.

70 citations

01 Jan 2016
TL;DR: The handbook of acoustical measurements and noise control is universally compatible with any devices to read and an online access to it is set as public so you can download it instantly.
Abstract: handbook of acoustical measurements and noise control is available in our digital library an online access to it is set as public so you can download it instantly. Our digital library saves in multiple countries, allowing you to get the most less latency time to download any of our books like this one. Merely said, the handbook of acoustical measurements and noise control is universally compatible with any devices to read.

43 citations

References
More filters
Journal ArticleDOI
TL;DR: The thermal conductivity and thermoelectric power of a single carbon nanotube were measured using a microfabricated suspended device and shows linear temperature dependence with a value of 80 microV/K at room temperature.
Abstract: The thermal conductivity and thermoelectric power of a single carbon nanotube were measured using a microfabricated suspended device. The observed thermal conductivity is more than 3000 W/K m at room temperature, which is 2 orders of magnitude higher than the estimation from previous experiments that used macroscopic mat samples. The temperature dependence of the thermal conductivity of nanotubes exhibits a peak at 320 K due to the onset of umklapp phonon scattering. The measured thermoelectric power shows linear temperature dependence with a value of 80 microV/K at room temperature.

3,166 citations


"Electronics Thermal Management in I..." refers background in this paper

  • ...During this period, the thermal community responded with several initiatives that improved TIMs [32]–[35], thermal metrology [36], [37], modeling [38], [39], and various building block technologies, such as microchannel heatsinks [40]–[44], ionic wind [45], heat pipes [46], [47], and thermoelectric coolers [48], [49]....

    [...]

01 Jan 2012

1,224 citations


"Electronics Thermal Management in I..." refers background in this paper

  • ...5 ZB in 2015, with a projection of 40 ZB for 2020 [5]....

    [...]

Journal ArticleDOI
TL;DR: The resulting discrete Boltzmann models are based on a kinetic representation of the fluid dynamics, hence the drawbacks in conventional higher-order hydrodynamic formulations can be avoided.
Abstract: We present in detail a theoretical framework for representing hydrodynamic systems through a systematic discretization of the Boltzmann kinetic equation. The work is an extension of a previously proposed formulation. Conventional lattice Boltzmann models can be shown to be directly derivable from this systematic approach. Furthermore, we provide here a clear and rigorous procedure for obtaining higher-order approximations to the continuum Boltzmann equation. The resulting macroscopic moment equations at each level of the systematic discretization give rise to the Navier–Stokes hydrodynamics and those beyond. In addition, theoretical indications to the order of accuracy requirements are given for each discrete approximation, for thermohydrodynamic systems, and for fluid systems involving long-range interactions. All these are important for complex and micro-scale flows and are missing in the conventional Navier–Stokes order descriptions. The resulting discrete Boltzmann models are based on a kinetic representation of the fluid dynamics, hence the drawbacks in conventional higher-order hydrodynamic formulations can be avoided.

914 citations


"Electronics Thermal Management in I..." refers methods in this paper

  • ...Other techniques such as the lattice Boltzmann method [65] could be used as well....

    [...]

Journal ArticleDOI
TL;DR: In this article, the effects of the channel size on the flow patterns and heat transfer and pressure drop performance are reviewed in small hydraulic diameter channels, and the fundamental questions related to the presence of nucleate boiling and characteristics of flow boiling in microchannels and minichannels in comparison to that in the conventional channel sizes (3 mm and above) are addressed.

840 citations

Journal ArticleDOI
Issam Mudawar1
TL;DR: This paper explores the recent research developments in high-heat-flux thermal management and demonstrates that, while different cooling options can be tailored to the specific needs of individual applications, system considerations always play a paramount role in determining the most suitable cooling scheme.
Abstract: This paper explores the recent research developments in high-heat-flux thermal management. Cooling schemes such as pool boiling, detachable heat sinks, channel flow boiling, microchannel and mini-channel heat sinks, jet-impingement, and sprays, are discussed and compared relative to heat dissipation potential, reliability, and packaging concerns. It is demonstrated that, while different cooling options can be tailored to the specific needs of individual applications, system considerations always play a paramount role in determining the most suitable cooling scheme. It is also shown that extensive fundamental electronic cooling knowledge has been amassed over the past two decades. Yet there is now a growing need for hardware innovations rather than perturbations to those fundamental studies. An example of these innovations is the cooling of military avionics, where research findings from the electronic cooling literature have made possible the development of a new generation of cooling hardware which promise order of magnitude increases in heat dissipation compared to today's cutting edge avionics cooling schemes.

824 citations


"Electronics Thermal Management in I..." refers background in this paper

  • ...During this period, the thermal community responded with several initiatives that improved TIMs [32]–[35], thermal metrology [36], [37], modeling [38], [39], and various building block technologies, such as microchannel heatsinks [40]–[44], ionic wind [45], heat pipes [46], [47], and thermoelectric coolers [48], [49]....

    [...]