ASME Conference Presenter Attendance Policy and Archival Proceedings

2015;():V001T00A001. doi:10.1115/IPACK2015-NS1.

This online compilation of papers from the ASME 2015 International Technical Conference and Exhibition on Packaging and Integration of Electronic and Photonic Microsystems (InterPACK2015) represents the archival version of the Conference Proceedings. According to ASME’s conference presenter attendance policy, if a paper is not presented at the Conference, the paper will not be published in the official archival Proceedings, which are registered with the Library of Congress and are submitted for abstracting and indexing. The paper also will not be published in The ASME Digital Collection and may not be cited as a published paper.

Commentary by Dr. Valentin Fuster

Thermal Management: Air Cooling: Heat Sink to System Level

2015;():V001T09A001. doi:10.1115/IPACK2015-48085.

When converting an electric power by an insulated-gate bipolar transistor (IGBT) module, the problem which is the heat generation in the IGBT module should be prudently considered in the design process. As an engineer reviews the cooling performance of power semi-conductor devices only at the component level, it is difficult to predict the reduction of airflow rates in the heat sink when power semi-conductor devices including the heat sink are integrated into the power conversion system. As the porous media model is adopted in the IGBT stack of the PCS, the problem that the meshes are heavily concentrated in the IGBT module including the heat sink, air, and IGBT/diode chips can be evaded and the airflow rate which is reflected in the effect of flow resistance by all interior structures including the IGBT module is calculated. For the outdoor type PCS, the hotspot temperature on the heat sink of the simulation and experiment is 99.3 and 101.6 Celsius, respectively. The proposed numerical simulation model considerably accurately predicts the hotspot temperature on the heat sink and can earn benefits in terms of efforts of mesh generation and computation time.

Commentary by Dr. Valentin Fuster
2015;():V001T09A002. doi:10.1115/IPACK2015-48151.

In this study, an oscillating-fan cooling device using electromagnetic force has been proposed. The device consists of two oscillating-fans flapping back and forth. It requires only one electromagnet and two elastic blades with one magnet on each of them. The electromagnet and two elastic blades are situated on a base and arranged accordingly. And thus, the electromagnetic force generated by the electromagnet can actuate the blades. The main advantage of this cooling device compared to a rotary fan is its simple structure because there is no bearing and motor in the cooling device. Moreover, the simplicity of the device makes it a highly reliable and low cost cooling device. The driving current can be either DC PWM or AC under 8 V – 12 V so it is compatible to most electronic devices. The dimensions of the cooling device can be designed as small as 20 mm (L) * 30 mm (W) * 4 mm (H) and as large as 60 mm * 55 mm * 25 mm. For a cooling experiment, three cooling devices with the dimension of 50 mm * 50 mm * 15 mm were incorporated with a heat sink with the dimension of 190 mm * 110 mm * 15 m. The dummy heater dissipated 55W while the environmental temperature is 44.8 °C. The result showed that the dummy heater can be cooled from 120.7 °C to 69.3 °C while the total power consumption of the three cooling devices is 1.74 W. The result shows that the cooling device not only provides an outstanding cooling ability but also shows a great potential for structural reliability and design flexibility.

Commentary by Dr. Valentin Fuster
2015;():V001T09A003. doi:10.1115/IPACK2015-48191.

Numerical study is carried out on natural convection heat transfer from three radial heat sinks subject to the influence of orientation. A finite volume method (FVM) numerical model was used to analyze the thermal performance of the radial heat sinks under upward, sideward and downward orientations. The effects of orientation with respect to gravity, fin number (15–30), the thickness of concentric ring (0.15–0.60) and Elenbaas number (15–55) on Nusselt number are investigated. Numerical results indicate that radiation is non-negligible in this study due to its high influence on thermal performance. The Nusselt number is relatively insensitive to the smaller ring thickness. The sideward facing orientation yields the worst thermal performance despite fin number changing. It is found that the thermal performance of heat sinks in upward and downward orientations depend on the number of fins significantly.

Commentary by Dr. Valentin Fuster
2015;():V001T09A004. doi:10.1115/IPACK2015-48511.

Heat transfer in a high aspect ratio, rectangular mm-scale channel that models a segment of a high-performance, air-cooled heat-sink is enhanced by deliberate formation of unsteady small-scale vortical motions. These small-scale motions are induced by self-fluttering, cantilevered planar thin-film reeds that are placed along the channel’s centerline. Heat transfer is enhanced by significant increases in both the local heat transfer coefficient at the fins surfaces, and in the mixing between the thermal boundary layers and the cooler core flow. The present investigation characterizes the thermal performance enhancement by reed actuation compared to the base flow (in the absence of the reeds) in terms of increased power dissipation over a range of flow rates, along with the associated fluid power. It is shown that because the cooling flow rate that is needed to sustain a given heat flux at a given surface temperature is almost two times higher than in the presence of the reeds, the reeds lead to a four-fold increase in thermal performance (as measured by the ratio of power dissipated to fluid power). The thermal effectiveness of the reeds is tested in a multi-channel heat sink, and it is shown that the improvement in heat transfer coefficient of the base flow is similar to that of the single channel.

Commentary by Dr. Valentin Fuster
2015;():V001T09A005. doi:10.1115/IPACK2015-48630.

A computational study is conducted to explore thermal performances of natural convection hybrid fin heat sinks (HF HSs). The proposed HF HSs are a hollow hybrid fin heat sink (HHF HS) and a solid hybrid fin heat sink (SHF HS). Parametric effects such as a fin spacing, an internal channel diameter, a heat dissipation on the performance of HF HSs are investigated by CFD analysis. Study results show that the thermal resistance of the HS increases while the mass-multiplied thermal resistance of the HS decreases associated with the increase of the channel diameter. The results also shows the thermal resistance of the SHF HS is 13% smaller, and the mass-multiplied thermal resistance of the HHF HS is 32% smaller compared with the pin fin heat sink (PF HS). These interesting results are mainly due to integrated effects of the mass-reduction, the surface area enhancement, and the heat pumping via the internal channel. Such better performances of HF HSs show the feasibility of alternatives to the conventional PF HS especially for passive cooling of LED lighting modules.

Commentary by Dr. Valentin Fuster
2015;():V001T09A006. doi:10.1115/IPACK2015-48641.

The thermal performance of heat sinks is commonly measured using heat sources with spring loaded thermocouples contained within plastic poppets that press against the heat sink to measure its surface temperature where the heat is applied. However, when the thickness of the heat sink base is small or the effective heat transfer coefficient on the fin side is large, the temperature at the thermocouple contact point is less than the nearby temperature where the heat source contacts the heat sink. This temperature depression under the contact thermocouples has been studied. The heat conduction equation is solved analytically to determine the temperature distribution around the contact thermocouple using a one-dimensional approximation and also a detailed two-dimensional approach. Two dimensionless groups are identified that characterize the detailed two-dimensional solution. The combination of the two dimensionless groups also appears in the one dimensional solution. The temperature distributions are validated using finite difference numerical solutions. It is shown that the one dimensional solution is the limit of the detailed solution when one of the dimensionless groups tends to infinity. A simple equation is provided to estimate the temperature measurement error on the heat sink surface.

Commentary by Dr. Valentin Fuster
2015;():V001T09A007. doi:10.1115/IPACK2015-48671.

Ever-increasing both data speed and traffic volume in the network telecommunications; as a result, producing more heat loss, challenges the conventional cooling methods. An optical plug module is a transceiver in data communication applications. By increasing the cooling demands, new thermal management solutions are necessary for optical plug modules. This article experimentally studies the heat pipe based cooling solutions for the optical plug modules. Heat pipes can passively transfer part of the produced heat from the hardly accessible places of the modules and expose it to the present active air cooling. Three different heat pipe based arrangements for a four-port optical plug assembly at both free and forced convection were investigated. Based on the results heat pipes helped to reduce heat sinks and total thermal resistance of this assembly on average by 27% and 16%, respectively under airflow rate of 10 ft3/min.

Commentary by Dr. Valentin Fuster
2015;():V001T09A008. doi:10.1115/IPACK2015-48745.

Metal foams are structures of a cellular nature that contain a high percentage of porosity that can be produced in either a closed or open cell forms. The use of metal foams in engineering applications has increased significantly over the last decade due to their enhanced mechanical and thermal properties. An innovative approach for three-dimensional (3D) detailed finite element modeling of open cell metal foam has been taken to capture the versatile nature of metal foams’ geometry and predicting its thermal performance. The interior complex geometry of metal foams has limited studies to create computational model via common approaches. Therefore, not much computational work has been done in open cell metal foam applications. To overcome this difficulty, computed tomography (CT) scan has been used to extract the 3D structure surface model with extreme precision. Computer-Aided Design (CAD) software has been used for “stitching” and “healing” of the CT scan model before importing it to a finite element domain. The 3D computational model is used in a heat sink application and is calibrated against experimental results for the temperature distribution of one case. The validated and calibrated model is then used for simulating different metal foam heat sink cases to assess the thermal and mechanical behavior under different conditions.

Commentary by Dr. Valentin Fuster
2015;():V001T09A009. doi:10.1115/IPACK2015-48753.

In this experiment, agitation is used in a rectangular channel for convective heat transfer enhancement. The channel under study is representative of a flow channel in an electronics cooling finned heat sink module. It is open at one end and has a translationally oscillating plate within it that agitates the flow. Contrary to the heat sink cooling channel, the test channel has no net through-flow so that agitation, isolated from throughflow effects, is studied. The channel is divided into three regions. The entry region is close to the open end of the channel. This would be near the fin tips in the finned heat exchanger channel. The base region is close to the other end of the channel where the flow makes an abrupt U-bend around the agitator plate. This is near the fin base region of a finned channel of a heat sink heat exchanger. The central region is between the two. Each region has special flow and convective heat transfer features for study. Ensemble-averaged velocities and RMS fluctuations of velocity are measured over the cycle. Measured data lend insight into the mixing phenomena in each region over the oscillation cycle. Unsteady heat flux measurements were made in each region and over the cycle to help in understanding the mechanisms affecting heat transfer. The unsteady heat flux characteristics in the entry and base regions seem to be more influenced by the RMS fluctuations of velocity, indicating that heat transfer in these regions is governed by turbulence generated by agitation. The unsteady heat flux trends in the central region seem to be more influenced by acceleration/deceleration of the flow than by turbulence-like structures.

Commentary by Dr. Valentin Fuster

Thermal Management: Data Centers and Energy Efficient Electronic Systems

2015;():V001T09A010. doi:10.1115/IPACK2015-48015.

In recent years, the internet services industry has been developing rapidly. Accordingly, the demands for compute and storage capacity continue to increase and internet data centers are consuming more power than ever before to provide this capacity. Based on the Forest & Sullivan market survey, data centers across the globe now consume around 100GWh of power and this consumption is expected to increase 30% by 2016. With development expanding, IDC (Internet Data Center) owners realize that small improvements in efficiency, from architecture design to daily operations, will yield large cost reduction benefits over time.

Cooling energy is a significant part of the daily operational expense of an IDC. One trend in this industry is to raise the operational temperature of an IDC, which also means running IT equipment at HTA (Higher Ambient Temperature) environment. This might also include cooling improvements such as water-side or air-side economizers which can be used in place of traditional closed loop CRAC (Computer room air conditioner) systems. But just raising the ambient inlet air temperature cannot be done by itself without looking at more effective ways of managing cooling control and considering the thermal safety.

An important trend seen in industry today is customized design for IT (Information Technology) equipment and IDC infrastructure from the cloud service provider. This trend brings an opportunity to consider IT and IDC together when designing and IDC, from the early design phase to the daily operation phase, when facing the challenge of improving efficiency. This trend also provides a chance to get more potential benefit out of higher operational temperature. The advantages and key components that make up a customized rack server design include reduced power consumption, more thermal margin with less fan power, and accurate thermal monitoring, etc. Accordingly, the specific IDC infrastructure can be re-designed to meet high temperature operations.

To raise the supply air temperature always means less thermal headroom for IT equipment. IDC operators will have less responses time with large power variations or any IDC failures happen. This paper introduces a new solution called ODC (on-demand cooling) with PTAS (Power Thermal Aware Solution) technology to deal with these challenges. ODC solution use the real time thermal data of IT equipment as the key input data for the cooling controls versus traditional ceiling installed sensors. It helps to improve the cooling control accuracy, decrease the response time and reduce temperature variation. By establishing a smart thermal operation with characteristics like direct feedback, accurate control and quick response, HTA can safely be achieved with confidence. The results of real demo testing show that, with real time thermal information, temperature oscillation and response time can be reduced effectively.

Topics: Data centers
Commentary by Dr. Valentin Fuster
2015;():V001T09A011. doi:10.1115/IPACK2015-48069.

The heat dissipated by electronic equipment inside data centers is increasing at a rapid rate due to the increasing of performance requirement and package density. This ever increasing power leads to critical challenges of thermal management for these high power density data centers. Energy consumption is also a key issue for high density data centers. Roughly 1.5% of all U.S. electricity consumption in the year 2006 was related to data centers, while that number increased to 2% by the year 2010. In 2013, U.S. data centers consumed approximately 91 billion kilowatt-hours of electricity. This amount of the electricity equals the annual output of 34 500-megawatt coal-fired power plants [1]. Cooling systems constitute a significant portion of the energy consumption of data centers, being approximately 25%∼35% of the total energy usage. Therefore, there is a large potential to save energy by optimizing current existing cooling systems and investigating new cooling technologies, and, at the same time, improving the overall cooling capacity and efficiency. This paper describes and investigates a hybrid cooling technology which utilizes in row coolers in existing raised floor air cooled data centers. The in row cooler functions as a liquid-to-air heat exchanger. In addition to the traditional raised floor cold aisle-hot aisle arrangements, the in row cooler is installed between the IT equipment to enable delivering the liquid coolant medium closer to the IT equipment. The in row coolers intake the hot air from the hot aisle, condition it, and supply the chilled air to the cold aisle. Thus, by extracting a large portion of the heat more directly into the cooling liquid through the in row coolers compared with the perimeter CRAH unit, the overall cooling performance and efficiency can potentially be improved. CFD models for an in row cooler and a representative data center room are developed. Experimentally characterized performance data are used to calibrate and validate the models. The models are then used to conduct a detailed computational analysis to assess the effectiveness of different arrangement configurations of in row cooler units in two rows of racks along one cold aisle. The detailed performance of the entire cold aisle is characterized using the rack inlet air temperature and a temperature nonuniformity factor. The impact of CRAH location and room layout are also investigated. This study is based on a practical problem and the corresponding results and analysis provide basic installation and design guidelines for future equipment upgrading in certain parts of the data center.

Topics: Cooling , Data centers
Commentary by Dr. Valentin Fuster
2015;():V001T09A012. doi:10.1115/IPACK2015-48071.

The heat dissipated by high performance IT equipment such as servers and switches in data centers is increasing rapidly, which makes the thermal management even more challenging. IT equipment is typically designed to operate at a rack inlet air temperature ranging between 10 °C and 35 °C. The newest published environmental standards for operating IT equipment proposed by ASHARE specify a long term recommended dry bulb IT air inlet temperature range as 18°C to 27°C. In terms of the short term specification, the largest allowable inlet temperature range to operate at is between 5°C and 45°C. Failure in maintaining these specifications will lead to significantly detrimental impacts to the performance and reliability of these electronic devices. Thus, understanding the cooling system is of paramount importance for the design and operation of data centers. In this paper, a hybrid cooling system is numerically modeled and investigated. The numerical modeling is conducted using a commercial computational fluid dynamics (CFD) code. The hybrid cooling strategy is specified by mounting the in row cooling units between the server racks to assist the raised floor air cooling. The effect of several input variables, including rack heat load and heat density, rack air flow rate, in row cooling unit operating cooling fluid flow rate and temperature, in row coil effectiveness, centralized cooling unit supply air flow rate, non-uniformity in rack heat load, and raised floor height are studied parametrically. Their detailed effects on the rack inlet air temperatures and the in row cooler performance are presented. The modeling results and corresponding analyses are used to develop general installation and operation guidance for the in row cooler strategy of a data center.

Commentary by Dr. Valentin Fuster
2015;():V001T09A013. doi:10.1115/IPACK2015-48152.

The generation-to-generation IT performance and density demands continue to drive innovation in data center cooling technologies. For many applications, the ability to efficiently deliver cooling via traditional chilled air cooling approaches has become inadequate. Water cooling has been used in data centers for more than 50 years to improve heat dissipation, boost performance and increase efficiency. While water cooling can undoubtedly have a higher initial capital cost, water cooling can be very cost effective when looking at the true lifecycle cost of a water cooled data center.

This study aims at addressing how one should evaluate the true total cost of ownership for water cooled data centers by considering the combined capital and operational cost for both the IT systems and the data center facility. It compares several metrics, including return-on-investment for three cooling technologies: traditional air cooling, rack-level cooling using rear door heat exchangers and direct water cooling via cold plates. The results highlight several important variables, namely, IT power, data center location, site electric utility cost, and construction costs and how each of these influence the total cost of ownership of water cooling. The study further looks at implementing water cooling as part of a new data center construction project versus a retrofit or upgrade into an existing data center facility.

Topics: Cooling , Data centers , Water
Commentary by Dr. Valentin Fuster
2015;():V001T09A014. doi:10.1115/IPACK2015-48169.

The advent of the big data era, the rapid development of mobile internet, and the rising demand of cloud computing services require increasingly more compute capability from their data center. This compute increase will most likely come from higher rack and room power densities or even construction of new Internet data centers. But an increase in a data center’s business-critical IT equipment (servers, hubs, routers, wiring patch panels, and other network appliances), not to mention the infrastructure needed to keep these devices alive and protected, encroaches on another IT goal: to reduce long-term energy usage. Large Internet Data Centers are looking at every possible way to reduce the cooling cost and improve efficiency. One of the emerging trends in the industry is to move to higher ambient data center operation and use air side economizers. However, these two trends can have significant implications for corrosion risk in data centers.

The prevailing practice surrounding the data centers has often been “The colder, the better.” However, some leading server manufacturers and data center efficiency experts share the opinion that data centers can run far hotter than they do today without sacrificing uptime, and with a huge savings in both cooling related costs and CO2 emissions. Why do we need to increase the temperatures? To cool data center requires huge refrigeration system which is energy hog and also cost of cooling infrastructure, maintenance cost and operation cost are heavy cost burden.

Ahuja et al [1] studied cooling path management in data center at typical operating temperature as well as higher ambient operating temperatures. High Temperatures and Corrosion Resistance technology will reduce the refrigeration output and how this innovation will open up new direction in data centers.

Note that, HTA is not to say that the higher the better. Before embracing HTA two keys points need to be addressed and understood. Firstly, server stability along with optimal temperature from data center perspective. Secondly, corrosion resistant technology. With Fresh air cooling the server has to bear with the seasons and diurnal variation of temperatures and these can be over 35 degree C, therefore to some extent, we have to say, HTA design is the premise of corrosion resistant design.

In this paper, we present methods to realize precise HTA operation along with corrosive resistant technology. This is achieved through an orchestrated collaboration between the IT and cooling infrastructures.

Commentary by Dr. Valentin Fuster
2015;():V001T09A015. doi:10.1115/IPACK2015-48176.

The energy used by information technology (IT) equipment and the supporting data center equipment keeps rising as data center proliferation continues unabated. In order to contain the rising computing costs, data center administrators are resorting to cost cutting measures such as not tightly controlling the temperature and humidity levels and in many cases installing air side economizers with the associated risk of introducing particulate and gaseous contaminations into their data centers. The ASHRAE TC9.9 subcommittee, on Mission Critical Facilities, Data Centers, Technology Spaces, and Electronic Equipment, has accommodated the data center administrators by allowing short period excursions outside the recommended temperature-humidity range, into allowable classes A1-A3. Under worst case conditions, the ASHRAE A3 envelope allows electronic equipment to operate at temperature and humidity as high as 24°C and 85% relative humidity for short, but undefined periods of time. This paper addresses the IT equipment reliability issues arising from operation in high humidity and high temperature conditions, with particular attention paid to the question of whether it is possible to determine the all-encompassing x-factors that can capture the effects of temperature and relative humidity on equipment reliability. The role of particulate and gaseous contamination and the aggravating effects of high temperature and high relative humidity will be presented and discussed. A method to determine the temperature and humidity x-factors, based on testing in experimental data centers located in polluted geographies, will be proposed.

Commentary by Dr. Valentin Fuster
2015;():V001T09A016. doi:10.1115/IPACK2015-48234.

As the number of data centers is exponentially growing globally, pragmatic characterization schemes are considered to be a necessity for measuring and modeling the load capacity and flow pattern of the facility. This paper contains experimental and numerical characterization of a new data center laboratory using practical measurements methods, including tiles and CRAH flow measurements. Then a full physics based CFD model is built to simulate/predict the measured data. A rapid flow curve method is used showing high accuracy and low computational expense.

Detailed descriptions of the data center structure, dimensions, layout (Appendix, A-1) and flow devices are given. Also, modeling parameters are mentioned in details to provide a baseline for any investigative parametric or sensitivity studies. Four experimental room level flow constraint scenarios are applied at which measurements were taken, (Appendix, A-2). The model is then built and calibrated then used to predict measurements.

Measurements of the cooling unit were performed using hot wire anemometry with a traverse duct installed at the top of the CRAH. The tiles measurements were carried out using a flow hood with back pressure compensation. A detailed CFD model is constructed to predict the four experimental cases. For modeling the interdependency between the flow and pressure in flow devices flow curve approach is used. This is a rapid modeling technique that relies on experimentally measured (for IT) or approximated (for CRAH) flow curves. Applying the operational flow curves boundary conditions at the vents of the flow device results in a very accurate simulation model. It is also shown that the flow curves can be used to predict the real-time flow rate of servers at known RPM. This greatly simplifies flow rate measurements of IT in the data center.

Commentary by Dr. Valentin Fuster
2015;():V001T09A017. doi:10.1115/IPACK2015-48237.

The perpetual increase of data processing has led to an ever increasing need for power and in turn to greater cooling challenges. High density (HD) IT loads have necessitated more aggressive and direct approaches of cooling as opposed to the legacy approach by the utilization of row-based cooling. In-row cooler systems are placed between the racks aligned with row orientation; they offer cool air to the IT equipment more directly and effectively. Following a horizontal airflow pattern and typically occupying 50% of a rack’s width; in-row cooling can be the main source of cooling in the data center or can work jointly with perimeter cooling. Another important development is the use of containment systems since they reduce mixing of hot and cold air in the facility. Both in-row technology and containment can be combined to form a very effective cooling solution for HD data centers.

This current study numerically investigates the behavior of in-row coolers in cold aisle containment (CAC) vs. perimeter cooling scheme. Also, we address the steady state performance for both systems, this includes manufacturer’s specifications such as heat exchanger performance and cooling coil capacity.

A brief failure scenario is then run, and duration of ride through time in the case of row-based cooling system failure is compared to raised floor perimeter cooling with containment. Non-raised floor cooling schemes will reduce the air volumetric storage of the whole facility (in this small data center cell it is about a 20% reduction). Also, the varying thermal inertia between the typical in-row and perimeter cooling units is of decisive importance.

The CFD model is validated using a new data center laboratory at Binghamton University with perimeter cooling. This data center consists of one main Liebert cooling unit, 46 perforated tiles with 22% open area, 40 racks distributed on three main cold aisles C and D. A computational slice is taken of the data center to generalize results. Cold aisle C consists of 16 rack and 18 perforated tiles with containment installed. In-row coolers are then added to the CFD model. Fixed IT load is maintained throughout the simulation and steady state comparisons are built between the legacy and row-based cooling schemes. An empirically obtained flow curve method is used to capture the flow-pressure correlation for flow devices.

Performance scenarios were parametrically analyzed for the following cases: (a) Perimeter cooling in CAC, (b) In-row cooling in CAC. Results showed that in-row coolers increased the efficiency of supply air flow utilization since the floor leakage was eliminated, and higher pressure build up in CAC were observed. This reduced the rack recirculation when compared to the perimeter cooled case. However, the heat exchanger size demonstrated the limitation of the in-row to maintain controlled set point at increased air flow conditions. For the pump failure scenario, experimental data provided by Emerson labs were used to capture the thermal inertia effect of the cooling coils for in-row and perimeter unit, perimeter cooled system proved to have longer ride through time.

Commentary by Dr. Valentin Fuster
2015;():V001T09A018. doi:10.1115/IPACK2015-48274.

Data center energy consumption can be divided into three broad categories: Information Technology (IT), Electrical, and Mechanical. An efficient data center uses the least amount of non-IT energy, which is typically divided between the mechanical and electrical systems. Mechanical systems generally contribute a large portion of the non-IT energy use by providing cooling from compressor-based equipment [1,2] and because of this, strategies to reduce compressor energy consumption can lead to significant mechanical system energy savings. The most efficient way to reduce compressor energy is through elimination or significant reduction in annual runtime. This is possible with the use of integrated airside or waterside economizers.

This paper demonstrates the impacts of economization in data centers through data collected from four operating facilities over the course of implementing various economizer improvement projects. System architectures include water-cooled centrifugal chiller plant with waterside economization, direct expansion air handling units (AHU) with airside economization, air-cooled centrifugal chillers with integrated waterside economization, and direct expansion computer room air conditioners (CRAC) with evaporative cooling and waterside economization. A systematic and methodical comparison of the baseline and post-conditions is discussed, comparing expected to observed economizer operating conditions. The comparison of multiple real-world scenarios revealed a range of variances in expected operation of economizer sequences to actual observations, indicating a need for close monitoring of system performance by data center operators to fully realize economizer benefits within facilities.

Topics: Data centers
Commentary by Dr. Valentin Fuster
2015;():V001T09A019. doi:10.1115/IPACK2015-48349.

Many data center operators are considering the option to convert from mechanical to free air cooling to improve energy efficiency. The main advantage of free air cooling is the elimination of chiller and Air Conditioning Unit operation when outdoor temperature falls below the data center temperature setpoint. Accidental introduction of gaseous pollutants in the data center along the fresh air and potential latency in response of control infrastructure to extreme events are some of the main concerns for adopting outside air cooling in data centers. Recent developments of ultra-high sensitivity corrosion sensors enable the real time monitoring of air quality and thus allow a better understanding of how airflow, relative humidity, and temperature fluctuations affect corrosion rates. Both the sensitivity of sensors and wireless networks ability to detect and react rapidly to any contamination event make them reliable tools to prevent corrosion related failures. A feasibility study is presented for eight legacy data centers that are evaluated to implement free air cooling.

Commentary by Dr. Valentin Fuster
2015;():V001T09A020. doi:10.1115/IPACK2015-48364.

Data centers are most commonly cooled by air delivered to electronic equipment from centralized cooling systems. The research presented here is motivated by the need for strategies to improve and optimize the load capacity and thermal efficiency of data centers by using computational fluid dynamics (CFD). Here, CFD is used to model and optimize the Villanova Steel Orca Research Center (VSORC). VSORC, presently in the design stages, will provide a testing environment as well as the capability to investigate best practices and state of the art strategies including hybrid cooling, IT load distribution, density zones, and hot aisle and cold aisle containment. The results of this study will be used in the overall design and construction of the aforementioned research data center. The objective of this study is to find the optimal operating points and design layout of a data center while still meeting certain design constraints. A focus is on finding both the ideal total supply flow rate of the air conditioning units and the ideal chilled water supply temperature (CHWST) setpoint under different data center design configurations and load capacities. The total supply flow rate of the air conditioning units and the supply temperature setpoint of the chilled water system are varied as design parameters in order to systematically determine the optimal operating points. The study also examines the influence of hot aisle and cold aisle containment strategies in full containment, half containment, and no containment configurations on the determined optimal operating conditions for the modeled research data center.

Commentary by Dr. Valentin Fuster
2015;():V001T09A021. doi:10.1115/IPACK2015-48375.

Data Center hybrid air/liquid cooling systems such as rear door heat exchangers, overhead and in row cooling systems enable localized, on-demand cooling, or “smart cooling.” At the heart of all hybrid cooling systems is an air to liquid cross flow heat exchanger that regulates the amount of cooling delivered by the system by modulating the liquid or air flows and/or temperatures. Due the central role that the heat exchanger plays in the system response, understanding the transient response of the heat exchanger is crucial for the precise control of hybrid cooling system. This paper reports on the transient experimental characterization of heat exchangers used in data centers applications. An experimental rig designed to introduce controlled transient perturbations in temperature and flow on the inlet air and liquid flow streams of a 12 in. × 12 in. heat exchanger test core is discussed. The conditioned air is delivered to the test core by a suction wind tunnel with upstream air heaters and a frequency variable axial blower to allow the control of air flow rate and bulk temperature. The conditioned water is delivered to the test core by a water delivery system consisting of two separate water circuits, one delivering cold water, and the other hot water. By switching from one circuit to the other or mixing water from both circuits, the rig is capable of generating step, ramp and frequency perturbations in water temperature at constant flow or step, ramp or frequency perturbations in water flow at constant temperature or combinations of temperature and water flow perturbations. Experimental data are presented for a 12×12 heat exchanger core with a single liquid pass under different transient perturbations.

Commentary by Dr. Valentin Fuster
2015;():V001T09A022. doi:10.1115/IPACK2015-48400.

Data centers are rapidly growing in size and number and consume an increasing and also significant proportion of energy production. Yet, their mission critical nature means they are constructed and operated at almost any cost. The data center industry is becoming more aware of the need to manage energy alongside managing capacity and availability. A vital element — and one that cannot be ignored — is how well cooling is delivered to IT equipment. Monitoring allows you to understand how well you are currently operating and whether you are within acceptable bounds. Meanwhile, simulation of the airflow and heat transfer allows you to predict future performance and understand current and future cooling issues. The challenge with both approaches is that they provide large volumes of data and interpretation can become a challenging task. Consider two scenarios:

i. One configuration results in a lot of ‘hot spots’ in the data center, resulting in equipment permanently operating in the ASHRAE Allowable range for intake air temperature.

ii. A second configuration has one ‘hot spot’ where equipment is operating above the ASHRAE Allowable range, but the remainder is within the ASHRAE Recommended range.

Which is better? One solution is to use metrics to help understand the performance. However, existing metrics are not always well known and understood. The only metric that is currently in common use is Power Usage Effectiveness (PUE). While PUE is useful as a measure of data center cooling efficiency, it does not address the cool air delivery effectiveness within the IT space. However, high-level cooling delivery metrics such as Rack Cooling Index (RCI), Return Temperature Index (RTI), Supply Heat Index (SHI), and Return Heat Index (RHI) have been recognized for producing key information but are not as widely used as they could be. Other metrics have been developed that give more detailed understanding of delivery performance. Capture Index, or more specifically Cold Aisle Capture Index (CACI) and Hot Aisle Capture Index (HACI) provide a measure of whether cooling systems targeting specific equipment work effectively. Simulation also allows diagnostic performance ‘measurement’ with detailed indices such as Rack and Room Recirculation. Since data center airflow is complex, this paper uses case studies to show how using metrics provides a rapid insight into both performance and what might need to be addressed to improve and optimize performance.

Commentary by Dr. Valentin Fuster
2015;():V001T09A023. doi:10.1115/IPACK2015-48423.

The most common approach to air cooling of data centers involves the pressurization of the plenum beneath the raised floor and delivery of air flow to racks via perforated floor tiles. This cooling approach is thermodynamically inefficient due in large part to the pressure losses through the tiles. Furthermore, it is difficult to control flow at the aisle and rack level since the flow source is centralized rather than distributed. Distributed cooling systems are more closely coupled to the heat generating racks. In overhead cooling systems, one can distribute flow to distinct aisles by placing the air mover and water cooled heat exchanger directly above an aisle. Two arrangements are possible: (i.) placing the air mover and heat exchanger above the cold aisle and forcing downward flow of cooled air into the cold aisle (Overhead Downward Flow (ODF)), or (ii.) placing the air mover and heat exchanger above the hot aisle and forcing heated air upwards from the hot aisle through the water cooled heat exchanger (Overhead Upward Flow (OUF)). This study focuses on the steady and transient behavior of overhead cooling systems in both ODF and OUF configurations and compares their cooling effectiveness and energy efficiency. The flow and heat transfer inside the servers and heat exchangers are modeled using physics based approaches that result in differential equation based mathematical descriptions. These models are programmed in the MATLAB™ language and embedded within a CFD computational environment (using the commercial code FLUENT™) that computes the steady or instantaneous airflow distribution. The complete computational model is able to simulate the complete flow and thermal field in the airside, the instantaneous temperatures within and pressure drops through the servers, and the instantaneous temperatures within and pressure drops through the overhead cooling system. Instantaneous overall energy consumption (1st Law) and exergy destruction (2nd Law) were used to quantify overall energy efficiency and to identify inefficiencies within the two systems. The server cooling effectiveness, based on an effectiveness-NTU model for the servers, was used to assess the cooling effectiveness of the two overhead cooling approaches.

Commentary by Dr. Valentin Fuster
2015;():V001T09A024. doi:10.1115/IPACK2015-48424.

Fast Fluid Dynamics (FFD), which has its origins in video game and movie animation applications, promises faster solve times than traditional RANS (Reynolds-Averaged Navier Stokes) CFD, is relatively easy to code, and is particularly suited to parallelization. Further, FFD is capable of modeling all relevant airflow physics including momentum, buoyancy and frictional effects which are not included in a standard Potential Flow Model (PFM). The present study is a first attempt to formally evaluate FFD for data center applications in which perforated tile airflow is predicted utilizing two-dimensional plenum models. Comparisons are made to RANS CFD and Potential Flow Modeling (PFM) over a variety of data center configurations based on five basic data center layouts, most of which are based on actual data centers. Results are compared to experimental measurements for one scenario.

Commentary by Dr. Valentin Fuster
2015;():V001T09A025. doi:10.1115/IPACK2015-48425.

The efficient control of cooling for data centers is an issue of broad economic importance due to the significant energy consumption of data centers. Many solutions attempt to optimize the control of the cooling equipment with temperature, pressure, or airflow sensors. We propose a simulation-based approach to optimize the cooling energy consumption and show how this approach can be implemented with simple power-consumption models. We also provide a real-life case study to demonstrate how energy saving cooling setpoints can be found using calibrated simulations and smooth metamodels of the system.

Commentary by Dr. Valentin Fuster
2015;():V001T09A026. doi:10.1115/IPACK2015-48430.

Lumped capacitance models have been introduced to study transient thermal response of data centers. Chilled water interruption of a Computer Room Air Handling (CRAH) unit is one of several failure scenarios of data center cooling infrastructures. In such a scenario, predicting the transient thermal response of the CRAH unit depends requires the determination of the CRAH lumped capacitance model parameters: the thermal capacitance (thermal mass) and the time constant. In this paper, we propose an experimental methodology to extract sufficient information for the lumped capacitance modeling of CRAH units. The method requires measurements of inlet and exit air temperature, air flow rate and CRAH fan power. If the chilled water supply to a CRAH unit is intentionally interrupted in a data center with multiple redundant CRAH units, sufficient information to estimate the CRAH lumped capacitance parameters can be obtained without disturbing the data center operation.

Commentary by Dr. Valentin Fuster
2015;():V001T09A027. doi:10.1115/IPACK2015-48432.

Existing tools for analyzing raised-floor-plenum airflow distribution focus on the prediction of tile-by-tile airflow rates and therefore necessarily require the input of a specific floor layout. Such CFD-based tools are also typically expensive, potentially slow and are of limited availability. The present study describes a design tool which takes a somewhat more qualitative approach in which overall tile airflow uniformity and the breakdown of cooling airflow (e.g., through the perforated tiles, raised-floor leakage, and cable cutouts) are estimated from only high-level design information. The design tool’s predictions are based on both CFD simulations performed “off line” and an analytical flow network model and can be used at the concept stage in a construction project or as a learning tool to quickly demonstrate the effect of design tradeoffs on plenum airflow distribution. The applicability of the tool to a real facility is confirmed through measured data and the tool is used to investigate the effect of various design choices on perforated tile airflow uniformity and distribution.

Topics: Design
Commentary by Dr. Valentin Fuster
2015;():V001T09A028. doi:10.1115/IPACK2015-48439.

One way to model the thermodynamic efficiency of a data center is to perform a second-law analysis using exergy calculations of the individual components. The in-house data center modeling tool Villanova Thermodynamic Analysis of Systems (VTAS) was applied to determine the effect of direct liquid cooling with cold plates on data center efficiency. The effectiveness of the cold plate as a function of the heat output of the servers was also examined. VTAS was used to study a configuration of two rows with eight racks each. Each rack was assumed to have twelve servers, each producing 200 W of heat. In addition to the cold plates and servers, this configuration also included a computer room air handling (CRAH) unit, a chiller, and a cooling tower. Preliminary results show that data center second-law efficiency increases as the cold plate removes more heat than the CRAH unit. Specifically, when using cold plates to remove all of the heat from the servers, the overall data center exergy destruction was reduced by over 30% compared to a configuration with only a CRAH unit and between 12–18% compared to a configuration with hybrid liquid-air technologies, thus showing that configurations with direct liquid cooling are thermodynamically more favorable. Furthermore, the effectiveness of the cold plate increases as the heat output of each server increases, suggesting that cold plates are most effective when removing a greater amount of heat.

Topics: Data centers
Commentary by Dr. Valentin Fuster
2015;():V001T09A029. doi:10.1115/IPACK2015-48448.

Calibrating a CFD model against measured data is the first step to successfully utilizing this technology for change-management and the optimization of an existing data center. To date, there has been very little published on this calibration process; more focus has been placed on the use of CFD at the design stage and the development of modeling techniques and solvers. Further, few studies which feature comprehensive comparisons of CFD-predicted and measured data have been published for real data centers, and many that have, demonstrated only modest agreement at best. This study provides another such comparison — for a 7,400 ft2 (687 m2), 138-rack, raised-floor facility. The goals of the study are to benchmark the level of agreement that can be practically obtained and also to investigate the level of modeling detail required. Additionally, specific practical advice covering both CFD modeling and experimental measurements is provided. A plenum-only CFD model is compared to measured tile airflow rates and a room-model, which uses measured tile flow rates as boundary conditions, is compared to temperatures measured at each rack inlet. The level of agreement is among the best published to date and demonstrates that a CFD model can be adequately calibrated against measured data and is of value for ongoing data center operation.

Commentary by Dr. Valentin Fuster
2015;():V001T09A030. doi:10.1115/IPACK2015-48466.

Energy efficiency is an essential element of server design for high performance computers. Traditional HPC servers or nodes that are air cooled enable efficiency by using optimized system design elements that include efficient heat sink design for critical components such as CPUs, Memory, Networking and Disk Subsystems. In addition, airflow optimization is enabled via critical component placement decisions as well as fan and cooling algorithms that have an objective to optimize airflow and maximize system performance. Critical elements that cannot be avoided in traditional air cooled servers are computer center level management of both the airflow requirements and the exhaust heat flux of the servers. An alternative approach shown in this paper uses a novel water cooled design that enables both extreme energy efficiency for heat extraction of the server heat load and allows for lower device operating temperatures for the critical components. Experimental data documented in this paper illustrates the advantages of using non-chilled water to cool the server, allowing 85 to 90 percent of the server heat load to be extracted by water while allowing inlet water temperatures up to 45 degrees Celsius. A comparison is made of the energy consumption needed to cool the server components for both the air cooled and water cooled systems. The base system used for the comparison uses identical system electronics and firmware. The server thermal data shown in the paper include thermal behavior at idle, typical and maximum power consumption states for the server. The data documents the range of boundary conditions that can be tolerated for water cooled server solutions and the comparative advantages of using this technology.

Topics: Cooling , Water
Commentary by Dr. Valentin Fuster
2015;():V001T09A031. doi:10.1115/IPACK2015-48470.

High performance computing server racks are being engineered to contain significantly more processing capability within the same computer room footprint year after year. The processor density within a single rack is becoming high enough that traditional, inefficient air-cooling of servers is inadequate to sustain HPC workloads. Experiments that characterize the performance of a direct water-cooled server rack in an operating HPC facility are described in this paper. Performance of the rack is reported for a range of cooling water inlet temperatures, flow rates and workloads that include actual and worst-case synthetic benchmarks. Power and temperature measurements of all processors and memory components in the rack were made while extended benchmark tests were conducted throughout the range of cooling variables allowed within an operational HPC facility. Synthetic benchmark results were compared with those obtained on a single server of the same design that had been characterized thermodynamically. Neither actual nor synthetic benchmark performances were affected during the course of the experiments, varying less than 0.13 percent. Power consumption change in the rack was minimal for the entire excursion of coolant temperatures and flow rates. Establishing the characteristics of such a highly energy efficient server rack in situ is critical to determine how the technology might be integrated into an existing heterogeneous, hybrid cooled computing facility — i.e., a facility that includes some servers that are air cooled as well as some that are direct water cooled.

Topics: Stress , Server racks , Water
Commentary by Dr. Valentin Fuster
2015;():V001T09A032. doi:10.1115/IPACK2015-48476.

High performance datacenters that are being built and operated to ensure optimized compute density for high performance computing (HPC) workloads are constrained by the requirement to provide adequate cooling for the servers. Traditional methods of cooling dense high power servers using air cooling imposes a large cooling and power burden on datacenters. Airflow optimization of the datacenter is a constraint subject to a high energy penalty when dense power hungry racks each capable of consuming 30 to 40 kW are populated in a dense datacenter environment. The work documented using a simulation model (TileFlow) in this paper demonstrates the challenges associated with a standard air cooled approach in a HPC datacenter. Alternate cooling approaches to traditional air cooling are simulated as a comparison to traditional air cooling. These include models using a heat exchanger assisted rack cooling solution with conventional chilled water and, a direct to node cooling model simulated for the racks.

These three distinct data center models are simulated at varying workloads and the resulting data is presented for typical and maximal inlet temperatures to the racks. For each cooling solution an estimate of the energy spend for the servers is determined based on the estimated PUEs of the cooling solutions chosen.

Topics: Cooling , Data centers , Water
Commentary by Dr. Valentin Fuster
2015;():V001T09A033. doi:10.1115/IPACK2015-48593.

For redundancy, almost all mission-critical facilities such as data centers are fitted with more air condition units than required. These units are most of the time heavily underutilized, where the fans within the units are still consuming energy circulating air without actually providing cooling. In more modern facilities such fans are equipped with variable frequency drives, which can reduce substantially the energy consumption if proper controls are implemented. While there have several solutions for controlling and optimizing such variable frequency drive operated air conditioning units, control systems without variable frequency drives (discrete on/off ACU controls) have not been addressed thoroughly. In this paper, we present a practical, distributed and automatic control method for such discrete air conditioning units. The technique includes several safety features and is based on dense environmental sensing and events like hotspots or device failures. We discuss this approach by way of example of a case study.

Commentary by Dr. Valentin Fuster
2015;():V001T09A034. doi:10.1115/IPACK2015-48684.

This paper presents the development of a neural network model of the server temperature to be used in model-based control of a data center. Data centers provide the optimal environments for operation of servers and storage devices. Conventionally, computational fluid dynamics (CFD) has been used to model the dynamic and complex environment of the data center. However, the drawback of this approach is its computational inefficiency. The effects of changing a single input may take an entire day to compute. Thus the CFD model is not well suited for model-based control. Instead, we propose to use an artificial Neural Network (NN) model which predicts server temperatures in significantly less time. In addition, this NN model has the capability of learning the environment in the data center by adapting its parameters in real time based on sensor data continuously taken from the data center. This work discusses the current development of the neural network, work being done at the University of Texas at Arlington, to include modeling of transient conditions, or time related changes, using data generated in a test bed Data Center at SUNY Binghamton.

Commentary by Dr. Valentin Fuster

Thermal Management: Power-Performance-Reliability Co-Design for ICs

2015;():V001T09A035. doi:10.1115/IPACK2015-48344.

Three-dimensional integration (3D IC) is a new technology that shows great potential for high performance and energy efficiency. However past work has shown that 3D ICs suffer from serious thermal issues, and advanced cooling solutions such as micro-fluidic cooling are necessary to realize the true potential of these systems. The interactions between thermal, electrical and physical aspects of a 3D design with micro-fluidic cooling are substantial, and a comprehensive co-design approach to address them simultaneously is a must. Such co-design techniques are required throughout the design process, including during architectural design space exploration (DSE) in order to ensure that optimal design choices are not overlooked. In this paper we propose a DSE framework for 3D CPUs with micro-fluidic cooling that applies electro-thermal optimization techniques to the circuit layout and the heatsink design. By considering such physical optimization techniques we provide a more accurate view of a 3D architecture’s thermal and timing feasibility, as well as its performance and energy efficiency. Using our proposed thermo-electrical-physical co-design DSE framework we are able to improve performance by 1.54x and energy efficiency by 1.26x.

Commentary by Dr. Valentin Fuster
2015;():V001T09A036. doi:10.1115/IPACK2015-48354.

3D ICs with through-silicon vias (TSVs) can achieve high performance while exacerbating the problem of heat removal. This necessitates the use of more aggressive cooling solutions such as micropin-fin based fluidic cooling. However, micropin-fin cooling comes with overheads such as non-uniform cooling capacity along the flow direction and restriction on the position of TSVs to where pins exist. 3D gate and TSV placement approaches un-aware of these drawbacks may lead to detrimental effects and even infeasible chip design. In this paper, we present a hierarchical partitioning based algorithm for co-placing gates and TSVs to co-optimize the wire-length and in-layer temperature uniformity, given the logical level netlist and layer assignment of gates. Compared to the wire-length driven gate placement followed by a TSV legalization stage, our approach can achieve up to 75% and 25% reduction of in-layer temperature variation and peak temperature, respectively, with the cost of 13% increase in wire-length.

Commentary by Dr. Valentin Fuster
2015;():V001T09A037. doi:10.1115/IPACK2015-48386.

This paper presents an electrical-thermal-reliability co-design technique for TSV-based 3D-ICs. Although TSV-based 3D-IC shows significant electrical performance improvement compared to traditional 2D circuit, researchers have reported strong electromigration (EM) in TSVs, which is induced by the thermal mechanical stress and the local temperature hotspot. We argue that rather than addressing 3D-IC’s EM issue after the IC designing phase, the designer should be aware of the circuit’s thermal and EM properties during the IC designing phase. For example, one should be aware that the TSVs establish vertical heat conduction path thus changing the chip’s thermal profile and also produce significant thermal mechanical stress to the nearby TSVs, which deteriorates other TSV’s EM reliability. Therefore, the number and location of TSVs play a crucial role in deciding 3D-IC’s electrical performance, changing its thermal profile, and affecting its EM-reliability. We investigate the TSV placement problem, in order to improve 3D-IC’s electrical performance and enhance its thermal-mechanical reliability. We derive and validate simple but accurate thermal and EM models for 3D-IC, which replace the current employed time-consuming finite-element-method (FEM) based simulation. Based on these models, we propose a systematic optimization flow to solve this TSV placement problem. Results show that compared to conventional performance-centered technique, our design methodology achieves 3.24x longer EM-lifetime, with only 1% performance degradation.

Topics: Reliability , Design
Commentary by Dr. Valentin Fuster
2015;():V001T09A038. doi:10.1115/IPACK2015-48533.

The high heat flux and strong thermal coupling in the 3D ICs has limited the performance gains that would otherwise be feasible in 3D structures. The common practice of adopting worst-case design margins is in part responsible for this limitation since average-case performance would be limited by worst-case thermal design margins. The coupling between temperature and leakage power exacerbates this effect. However, worst-case thermal conditions are not the common state across the package at runtime. We argue for the co-design of the package, architecture, and power management based on the multi-physics interactions between temperature, power consumption and system performance. This approach suggests an adaptive architecture that accommodates the thermal coupling between layers and leads to increased energy efficiency over a wider operating voltage range and therefore higher performance.

In this paper, we target at a 3D multicore architecture where the cores reside on one die and the last level cache (LLC) resides on the other. The DRAM stack may be stacked on top of the package (e.g., 3D) or in the same package (e.g., 2.5D). We propose a novel adaptive cache structure — the constant performance model (CPM) cache — based on voltage adaptations to temperature variations. We construct a HSPICE model for the SRAM to explore the relationship between temperature, supply voltage, and the circuit delay in the context of the LLC. This model is used to investigate, characterize, and analyze the effect of the temperature-delay dependence of the SRAM LLC configuration on the system-level performance and energy efficiency. This analysis gives rise to an intelligent scheme for dynamic voltage regulation in the LLC cache that is sensitive to the temperature of the individual cache banks. Each cache bank is thermally coupled to the associated cores and thus is sensitive to the local core-level power management. We show that this local adaptation to the temperature-delay dependence leads to a significant power reduction in the LLC cache, and improvement of system energy efficiency computed as energy per instruction (EPI). We evaluate our approach using a cycle-level, full system simulation model of a 16-core x86 homogenous microarchitecture in 16nm technology that boots a full Linux operating system and executes application binaries. The advantages of the proposed adaptive LLC structure illustrate the potential of the co-design of the package, architecture, and power management in future 3D multicore architectures.

Commentary by Dr. Valentin Fuster
2015;():V001T09A039. doi:10.1115/IPACK2015-48690.

A major challenge in the implementation of evaporative two-phase liquid-cooled ICs with embedded fluid microchannels/cavities is the high pressure drops arising from evaporation-induced expansion and acceleration of the flowing two-phase fluid in small hydraulic diameters. Our ongoing research effort addresses this challenge by utilizing a novel hierarchical radially expanding channel networks with a central embedded inlet manifold and drainage at the periphery of the chip stack. This paper presents a qualitative description of the thermal design process that has been adopted for this radial cavity. The thermal design process first involves construction of a system-level pressure-thermal model for the radial cavity based on both fundamental experiments as well as numerical simulations performed on the building block structures of the final architecture. Finally, this system-level pressure-thermal model can be used to identify the design space and optimize the geometry to maximize thermal performance, while respecting design specifications. This design flow presents a good case study for electrical-thermal co-design of two-phase liquid cooled ICs.

Commentary by Dr. Valentin Fuster

Thermal Management: Rack Level Thermal Management

2015;():V001T09A040. doi:10.1115/IPACK2015-48024.

Rotating fans are widely utilized in thermal management applications and their accurate characterization has recently become even a more critical issue for thermofluids engineers. The present study investigates the characterization of an axial fan computationally and experimentally. Using the three-dimensional CAD models of the fan, a series of computational fluid dynamics (CFD) simulations were performed to determine the flow and pressure fields produced by the axial mover over a range of flow rates. In order to validate the computational model findings, experiments were conducted to obtain the pressure drop values at different flow rates in an AMCA (Air Movement and Control Association) standard 210-99, 1999 wind tunnel. These data sets were also compared with the fan vendor’s published testing data. A reasonably good agreement was obtained among the data from these three separate sources. Furthermore, an attempt was made to understand the overall fan efficiency as a function of the volumetric flow rate. It was determined that the maximum overall fan efficiency was less than 27% correlating well with the computational results.

Commentary by Dr. Valentin Fuster
2015;():V001T09A041. doi:10.1115/IPACK2015-48244.

Because of the rapid growth in the number of data centers combined with the high density heat dissipation in the IT and telecommunications equipment, energy efficient thermal management of data centers has become a key research focus in the electronics packaging community. Traditional legacy data centers still rely largely on chilled air flow delivered to the IT equipment racks through perforated tiles from the raised floor plenum. When there is large variation in the amount of heat dissipated by the racks in a given aisle, the standard air cooling approach requires over-provisioning.

Localized hybrid air-water cooling is one approach to more effectively control the cooling when there is wide variation in the amount of dissipation in neighboring racks. In a closed hybrid air-water cooled server cabinet, the generated heat is removed by a self-contained system that does not interact with the room level air cooling system. In this study, a comprehensive procedure for CFD validation in a close coupled hybrid cooled enclosed cabinet is described. The commercial enclosure has been characterized experimentally in an earlier study, where the effectiveness values were applied as boundary conditions to the compact heat exchanger model.

Here, the previously obtained experimental data are used to validate the results from computational modeling. Two cases with different air flow rates are compared. Very good agreement is achieved, with the maximum overall average error less than 4%. Due to relatively high pressure inside the cabinet, it is possible that air leakage from the cabinet may be responsible for the discrepancy between the model and experimental results. A sensitivity study was applied to the validated model to investigate the effect leakage had on the cabinet’s performance.

Commentary by Dr. Valentin Fuster
2015;():V001T09A042. doi:10.1115/IPACK2015-48258.

With the advent of big data and cloud computing solutions, enterprise demand for servers is increasing. There is especially high growth for Intel based x86 server platforms. Today’s datacenters are in constant pursuit of high performance/high availability computing solutions coupled with low power consumption and low heat generation and the ability to manage all of this through advanced telemetry data gathering. This paper showcases one such solution of an updated rack and server architecture that promises such improvements.

The ability to manage server and data center power consumption and cooling more completely is critical in effectively managing datacenter costs and reducing the PUE in the data center. Traditional Intel based 1U and 2U form factor servers have existed in the data center for decades. These general purpose x86 server designs by the major OEM’s are, for all practical purposes, very similar in their power consumption and thermal output. Power supplies and thermal designs for server in the past have not been optimized for high efficiency. In addition, IT managers need to know more information about servers in order to optimize data center cooling and power use, an improved server/rack design needs to be built to take advantage of more efficient power supplies or PDU’s and more efficient means of cooling server compute resources than from traditional internal server fans. This is the constant pursuit of corporations looking at new ways to improving efficiency and gaining a competitive advantage.

A new way to optimize power consumption and improve cooling is a complete redesign of the traditional server rack. Extracting internal server power supplies and server fans and centralizing these within the rack aims to achieve this goal. This type of design achieves an entirely new low power target by utilizing centralized, high efficiency PDU’s that power all servers within the rack. Cooling is improved by also utilizing large efficient rack based fans for airflow to all servers. Also, opening up the server design is to allow greater airflow across server components for improved cooling. This centralized power supply breaks through the traditional server power limits. Rack based PDU’s can adjust the power efficiency to a more optimum point. Combine this with the use of online + offline modes within one single power supply. Cold backup makes data center power to achieve optimal power efficiency. In addition, unifying the mechanical structure and thermal definitions within the rack solution for server cooling and PSU information allows IT to collect all server power and thermal information centrally for improved ease in analyzing and processing.

Topics: Data centers
Commentary by Dr. Valentin Fuster
2015;():V001T09A043. doi:10.1115/IPACK2015-48377.

The cross flow heat exchanger is at the heart of most cooling systems for data centers. Air/Water or air/refrigerant heat exchangers are the principal component in Central Room Air Conditioning (CRAC) units that condition data room air that is delivered through an underfloor plenum. Liquid/air heat exchangers are also increasingly deployed in close-coupled cooling systems such as rear door heat exchangers, in-row coolers, and overhead coolers. In all cases, the performance of liquid/air heat exchangers in both steady state and transient scenarios are of principal concern. Transient scenarios occur either by the accidental failure of the cooling system or by intentional dynamic control of the cooling system. In either scenario, transient boundary conditions involve time-dependent air or liquid inlet temperatures and mass flow rates that may be coupled in any number of potential combinations. Understanding and characterizing the performance of the heat exchanger in these transient scenarios is of paramount importance for designing better thermal solutions and improving the operational efficiency of existing cooling systems. In this paper, the transient performance of water to air cross flow heat exchangers is studied using numerical modeling and experimental measurements. Experimental measurements in 12 in. × 12 in. heat exchanger cores were performed, in which the liquid (water) mass flow rate or inlet temperature are varied in time following controlled functional forms (step jump, ramp). The experimental data were used to validate a transient numerical model developed with traditional assumptions of space averaging of heat transfer coefficients, and volume averaging of thermal capacitances. The complete numerical model was combined with the transient effectiveness methodology in which the traditional heat exchanger effectiveness approach is extended into a transient domain, and is then used to model the heat exchanger transient response. Different transient scenarios were parametrically studied to develop an understanding of the impact of critical variables such as, the fluid inlet temperature variation and the fluid mass flow rate variation, and a more comprehensive understanding of the characteristics of the transient effectiveness. Agreement between the novel transient effectiveness modeling approach and the experimental measurements enable use of the models as verified predictive design tools. Several studies are designed based on the practical problems related to data center thermal environments and the results are analyzed.

Commentary by Dr. Valentin Fuster
2015;():V001T09A044. doi:10.1115/IPACK2015-48600.

This study describes an application of the flow resistance network analysis to thermal design of fan-cooled electronic equipment. Especially, a modeling method of the flow resistance network was investigated. Current electronic equipment becomes smaller and thinner while their functions become more complex. As a result, flow passages for cooling air become complex. In order to simulate the complex airflow in high-density packaging electronic equipment by using the flow resistance network, we tried to develop the flow resistance network by support of the 3D-CFD analysis. A test model which simulates high-density packaging electronic equipment is prepared and the flow resistance network analysis is applied to the prediction of flow rate distribution in the model. Through the investigation, we obtained information and future problems about the development of the flow resistance network in electronic equipment with lots of electrical components.

Commentary by Dr. Valentin Fuster
2015;():V001T09A045. doi:10.1115/IPACK2015-48607.

This study describes a possibility of an improvement of water cooling devices for high-power electronic devices such as inverters for electric vehicles by using a combination of micro heat sinks and miniature vortex generators. Power devices such as IGBT (Insulated Gate Bipolar Transistor) are widely used for controlling an operation of electronic vehicles and hybrid vehicles. Due to the improvement of the performance of the power devices, the heat dissipation density from these devices becomes higher. The water cooling is the commonest method for dissipating heat from the inverter of the electronic vehicles. Therefore the improvement of the water cooling technology is significantly needed in order to manage the increase of the heat dissipation density.

We are now trying to develop a high-performance water cooling device for dissipating the high heat flux from the inverters in the electric vehicles by using a combination of a fine miniature heat sink and a miniature vortex generator. The combination of the miniature heat sink and the vortex generator may increase heat transfer performance of the heat exchanger while inhibiting an increase of pressure drop by generating a swirling turbulent flow in a clearance between the heat sink fins. In this study, the water cooling performance in the narrow flow passage, which simulates the flow passage in the water cooling device, with the miniature heat sink and the miniature vortex generators was investigated by using 3-dimentional CFD analysis. From the analysis, we conclude that the combination of the miniature heat sink and the vortex generator was effective for the heat transfer enhancement in the narrow flow passage of the water cooling device while inhibiting the generation of the pressure drop when we can use the combination with the appropriate manner.

Commentary by Dr. Valentin Fuster
2015;():V001T09A046. doi:10.1115/IPACK2015-48640.

Efficient and compact cooling technologies play a pivotal role in determining the performance of high performance computing devices when used with highly parallel workloads in supercomputers. The present work deals with evaluation of different cooling technologies and elucidating their impact on the power, performance, and thermal management of Intel® Xeon Phi™ coprocessors. The scope of the study is to demonstrate enhanced cooling capabilities beyond today’s fan-driven air-cooling for use in high performance computing (HPC) technology, thereby improving the overall Performance per Watt in datacenters. The various cooling technologies evaluated for the present study include air-cooling, liquid-cooling and two-phase immersion-cooling. Air-cooling is evaluated by providing controlled airflow to a cluster of eight 300 W Xeon Phi coprocessors (7120P). For liquid-cooling, two different cold plate technologies are evaluated, viz, Formed tube cold pates and Microchannel based cold plates. Liquidcooling with water as working fluid, is evaluated on single Xeon Phi coprocessors, using inlet conditions in accordance with ASHRAE W2 and W3 class liquid cooled datacenter baselines. For immersion-cooling, a cluster of multiple Xeon Phi coprocessors is evaluated, with three different types of Integrated Heat Spreaders (IHS), viz., bare IHS, IHS with a Boiling Enhancement Coating (BEC) and IHS with BEC coated pin-fins. The entire cluster is immersed in a pool of Novec 649 (3M fluid, boiling point 49 °C at 1 atm), with polycarbonate spacers used to reduce the volume of fluid required, to achieve target fluid/power density of ∼ 3 L/kW. Flow visualization is performed to provide further insight into the boiling behavior during the immersion-cooling process.

Performance per Watt of the Xeon Phi coprocessors is characterized as a function of the cooling technologies using several HPC workloads benchmark run at constant frequency, such as the Intel proprietary Power Thermal Utility (PTU), and industry standard HPC benchmarks LINPACK, DGEMM, SGEMM and STREAM. The major parameters measured by sensors on the coprocessor include total power to the coprocessor, CPU temperature, and memory temperature, while the calculated outputs of interest also include the performance per watt and equivalent thermal resistance. As expected, it is observed that both liquid and immersion cooling show improved performance per Watt and lower CPU temperature compared to air-cooling. In addition to elucidating the performance/watt improvement, this work reports on the relationship of cooling technologies on total power consumed by the Xeon-Phi card as a function of coolant inlet temperatures. Further, the paper discusses form-factor advantages to liquid and immersion cooling and compares technologies on a common platform. Finally, the paper concludes by discussing datacenter optimization for cooling in the context of leakage power control for Xeon-Phi coprocessors.

Topics: Cooling
Commentary by Dr. Valentin Fuster
2015;():V001T09A047. doi:10.1115/IPACK2015-48771.

Liquid immersion cooling technology, currently in its nascence as a commercially available solution for data center installations, is growing in popularity as the power density of next-gen electronics necessitates a matriculation to thermal management techniques capable of handling incredibly high heat fluxes reliably and efficiently. The use of boiling and single-phase convective solutions using dielectric fluids can result in dramatic reductions in chip temperatures, thus increasing reliability. The latter method is growing in popularity faster than the former but, as both of these approaches gain acceptance, packaging engineers will require insight into how coolant is distributed throughout the enclosure for either solution. More specifically, analytical and experimental techniques will be required to ascertain how thermal performance and system efficiency of more critical elements, such as processor chips, are affected by the auxiliary components, heated or not, that must exist within a computing device. These supplemental components, whether entirely passive or modestly heated, if placed strategically can be integrated in such a way to improve the thermal performance of the system by guiding the coolant through the liquid filled enclosure. To this end, flow guides, which simulate these auxiliary components, have been integrated into a small form factor high performance server module. The relationship between the surface temperature and the power dissipated by the primary heated elements within the device has been explored as well as the pressure drop experienced by the coolant flowing through the enclosure. Power dissipations near 450W have been achieved at a surface temperature of approximately 75°C with the use of flow guides, a near 50W improvement over previous results. Furthermore, this value was attained at a modest pressure drop of 0.71 psi for the dielectric fluid flowing through the cartridge. Slightly over 300W of power dissipation was achieved at an even lower pressure drop of 0.13 psi at a similar operating temperature. Pool boiling results have shown that passive elements can have a significant impact on thermal performance. Reductions of nearly 50W in the maximum power dissipation achieved have been shown when the largest flow guide is integrated. A PIV analytical method is proposed and applied to the current experimental facility to assess the effectiveness of the flow guide design proposed.

Topics: Flow (Dynamics)
Commentary by Dr. Valentin Fuster

Thermal Management: Thermal Management: Device Level Thermal Management

2015;():V001T09A048. doi:10.1115/IPACK2015-48017.

In this paper, a centrifugal micropump was designed, fabricated and characterized. The proposed micropump is able to provide a 1.4L/min flow rate and a 75KPa pressure head at 24000 rpm with an oversize of 46mm wide and 69mm long. The hydrodynamic components were designed based on partial emission pump. Meanwhile, the geometric profiles of both impeller and volute were simplified for manufacturing. A computational fluid dynamics (CFD) analysis was performed to predict the effects of blade inlet angle and vane number on hydraulic performance. Experiments were conducted at 4 different rotational speeds to validate the numerical results. The results showed that the numerical simulation has a high accuracy to predict the micropump flow field with the overall average deviation less than 3%. As expected, the micropump prototype performed obvious partial emission pump features. In terms of the external characteristic, the pressure head at a given rotational speed decreased little with flow rate increasing. While, in the flow field, complex secondary flow was significant in the impeller passage, due to the joint action of the blade tip clearance leakage and axial vortex. Regression analysis and statistical evaluation showed that the flow nondimensional coefficients at different rotational speeds correlated well, indicating that classical similarity rules was still applicable to this micropump.

Commentary by Dr. Valentin Fuster
2015;():V001T09A049. doi:10.1115/IPACK2015-48019.

Heat pipes are recognized as an excellent heat transport devices and extensively investigated for applications in electronic cooling. Different types of heat pipes have been developed such as micro/miniature heat pipes, loop heat pipes and so on, and these heat pipes have been widely applied in the field of electronics cooling such as notebook, desktop, data center; as well as aerospace, industrial cooling field. However, in recent years the application of heat pipe is widening to the filed of hand held mobile electronic devices such as smart phone, tablet pc, digital camera etc. With the development in technology these devices have different user friendly functions and capabilities, which requires the highest processor clock speed. In general, high clock speed of processor generates lot of heat which need to be spread or removed to eliminate the hot spot. It becomes a challenging task to cool such electronic devices as mentioned above with a very confined space and concentrated heat sources. Regarding to this challenge, ultra thin flat heat pipe is developed; this newly developed heat pipe consists of a special fiber wick structure which can ensure vapor spaces on the two sides of the wick structure.

In this paper a novel thin spreader is proposed to eliminate the hot spot; generally the proposed heat spreader consists of 0.20mm thick metal plate and ultra thin heat pipe of 0.40mm thickness soldered in its body. Maximum thickness of this spreader is 0.63mm. Metal plate is 60mm × 110mm in size; and the ultra thin heat pipe can be fabricated from different original diameter ranges from 2.0mm to 3.0mm Cu tube. Theoretical and experimental analysis have been done to evaluate this thin spreader. In addition, some real application of this spreader will be introduced in this paper.

Topics: Flat heat pipes
Commentary by Dr. Valentin Fuster
2015;():V001T09A050. doi:10.1115/IPACK2015-48051.

In this paper, an analytical solution for the thermal behavior of rectangular flux channels with discretely specified boundary conditions is presented. The boundary conditions along the source plane can be a combination of contact temperatures, heat fluxes, and/or adiabatic. Convective cooling is applied along the sink plane, and the edges of the channel are assumed adiabatic. The governing equation of the system is the Laplace equation which is solved using the method of separation of variables and the least squares method. The solution is presented in the form of Fourier series expansion. As a case study, a symmetrical flux channel with a combination of five discretely specified boundary conditions, including temperature, heat flux and adiabatic conditions is considered. Temperature profile along the channel is calculated and compared with the Finite Element Method (FEM) using COMSOL commercial software package [1]. A good agreement is observed between the analytical and FEM results.

Topics: Temperature
Commentary by Dr. Valentin Fuster
2015;():V001T09A051. doi:10.1115/IPACK2015-48061.

A shell-and-tube phase change material (PCM) thermal energy storage (TES) unit has been analyzed numerically and experimentally. The tube bank is filled with commercial paraffin RUBITHERM RT 28 HC PCM, which melts as the heat transfer fluid (HTF) flows across the tubes. A fully-implicit one-dimensional control volume formulation that utilizes the enthalpy method for phase change has been developed to determine the transient temperature distributions in both the PCM and the tubes themselves. The energy gained by a column of tubes is used to determine the exit bulk HTF temperature from that column, ultimately leading to an exit HTF temperature from the TES unit. This paper presents a comparison of the numerical and experimental results for the transient temperature profiles of the PCM-filled tubes and HTF.

Commentary by Dr. Valentin Fuster
2015;():V001T09A052. doi:10.1115/IPACK2015-48119.

Three-dimensional finite-element modeling is used to determine the thermally optimum design of a GaN-on-SiC MMIC power amplifier, with a focus on the parametric influence of the thermal boundary resistance (TBR), epitaxial geometry, and dissipated linear power on the HEMT junction temperature rise. A commercial MMIC power amplifier is used to set the baseline geometry and dimensions. It is found that the frequently neglected Thermal Boundary Resistance (TBR), between the GaN and SiC, not only has a significant influence on the maximum junction temperature, but directly influences the thermally-optimal GaN thickness for the HEMT transistor. The thermally-optimal GaN thickness is a balance between spreading, vertical thermal resistance, and the magnitude of the TBR. As a consequence, it is seen the commonly used, submicron l GaN thicknesses approach optimality only when the TBR values are below 10 m2-K/GW. Additionally, it is observed that increasing the gate pitch and substrate thickness helps to diffuse the flow of heat within the substrate before it proceeds into the cooling solution, resulting in an overall decrease in thermal resistance. The numerical results are used to verify the accuracy of an available analytical solution for a surface heat source on an orthotropic multi-layer structure, albeit with assumed temperature-invariant properties, thus enabling use of this relation in scoping and preliminary design calculations.

Commentary by Dr. Valentin Fuster
2015;():V001T09A053. doi:10.1115/IPACK2015-48243.

Thermal spreading resistance in a multilayered orthotropic disk is considered. Interfacial resistance between each layer is prescribed by means of a contact conductance hc using a Robin type boundary condition. Orthotropic properties are considered by transforming the orthotropic system into an equivalent isotropic system using stretched coordinates. A recursive modeling approach is presented to account for the effects of two or more layers in the structure from the simple case of a single isotropic layer. This approach simplifies the analysis considerably. Finally, variable heat flux distribution is considered for three special cases: uniform, parabolic, and inverse parabolic. Numerous special cases can be derived from the general result including perfect interfacial contact and perfect sink plane conductance. Additional issues are also discussed in detail. The expressions for the total thermal resistance and spreading resistance can be easily implemented in any mathematical software or coded in Fortran, C, or BASIC. Since the method is strictly analytical, thermal analysts can quickly assess changes in layer properties, material sequence, heat flux distribution, and effects of interfacial contact resistance, with little extra effort.

Commentary by Dr. Valentin Fuster
2015;():V001T09A054. doi:10.1115/IPACK2015-48285.

In this research, operating characteristics and heat transfer phenomena in 2-turn pulsating heat pipe operating in a circulation mode were experimentally investigated. Temperature, pressure and high-speed flow visualization data were obtained with the variation of diameters (1.2 mm, 1.7 mm and 2.2 mm) and input powers. The overall pressure variation from start-up to steady state was measured using the pressure transmitters in the evaporator section. Heat transfer phenomena were investigated using homogeneous-equilibrium model. Thermodynamic state of two-phase mixture at the exit of evaporator is identified as a saturation state using obtained temperature and pressure data. The ratio of sensible heat to latent heat changed with the variation of diameters and input powers. It was found that each evaporator has a different ratio and latent heat was dominant in most experimental cases.

Topics: Fluids , Heat pipes
Commentary by Dr. Valentin Fuster
2015;():V001T09A055. doi:10.1115/IPACK2015-48333.

Thermal management problems in electronic packages have been a challenging problem due to increasing number of transistors in chips and reduction in product size. Thermal interface materials (TIM) help heat dissipation by reducing thermal contact resistance between chip and integrated heat spreader (IHS) and TIM quality is critical for effective removal of heat generated from the package. Therefore, identification of defects within TIM is required during package assembly process development. Imaging techniques such as computerized scanning acoustic microscopy (CSAM) and X-ray tomography are used as non-destructive testing techniques to identify TIM defects qualitatively. More recently, it was shown that IR thermography can be used as a qualitative means of identifying defects as well. Thermal diffusion tomography is a powerful alternative to those techniques due to its lower cost and ease of application. In this study, quantitative characterization of defects in TIM is presented using thermal diffusion tomography. The study is conducted considering a high density interconnect flip chip package that includes spreading effect due to different sized IHS and die. Defect size and location are detected analyzing the measured thermal response of electronic package by solving the resulting inverse problem by Levenberg-Marquardt algorithm as an image reconstruction technique.

Commentary by Dr. Valentin Fuster
2015;():V001T09A056. doi:10.1115/IPACK2015-48537.

An evaluation of two approaches to localized hotspot cooling is conducted through both numerical modeling and experimental demonstration, with the advantages and limitations of each approach highlighted. The first approach, locally increasing the density of pins in a micro pin fin heat sink, was shown through numerical modeling to deliver a factor of two enhancement in effective heat transfer coefficient by doubling the pin density near the hotspot. This simpler approach to maintaining temperature uniformity eliminates the need for hotspot specific fluid routing and delivery, and also has minimal impact on the larger flow field. Dedicated hotspot coolers, on the other hand, have the ability to dissipate significantly larger heat fluxes while maintaining manageable pressure drops, because the flow rate to the dedicated cooler can be closely matched to the demands of the hotspot. Dissipation of hotspot heat fluxes in excess of 2 kW/cm2 is demonstrated experimentally using a two phase dedicated hotspot cooler. However, dedicated coolers require additional fluidic routing and manifolding to efficiently deliver the coolant to the hotspot. These integration concerns are considered in concert with the performance of the hotspot cooler itself to enable better informed thermal design for both system level and device level cooling.

Topics: Fins , Coolers
Commentary by Dr. Valentin Fuster
2015;():V001T09A057. doi:10.1115/IPACK2015-48592.

Advances in manufacturing techniques are inspiring the design of novel integrated microscale thermal cooling devices seeking to push the limits of current thermal management solutions in high heat flux applications. These advanced cooling technologies can be used to improve the performance of high power density electronics such as GaN-based RF power amplifiers. However, their optimal design requires careful analysis of the combined effects of conduction and convection.

Many numerical simulations and optimization studies have been performed for single cell models of microchannel heat sinks, but these simulations do not provide insight into the flow and heat transfer through the entire device. This study therefore presents the results of conjugate heat transfer CFD simulations for a complex copper monolithic heat sink integrated with a 100 micron thick, 5 mm by 1 mm high power density GaN-SiC chip. The computational model (13 million cells) represents both the chip and the heat sink, which consists of multiple inlets and outlets for fluid entry and exit, delivery and collection manifold systems, and an array of fins that form rectangular microchannels. Total chip powers of up to 150 W at the GaN gates were considered, and a quarter of the device was modeled for total inlet mass flow rates of 1.44 g/s and 1.8 g/s (0.36 g/s and 0.45 g/s for the quarter device), corresponding to laminar flow at Reynolds numbers between 19.5 and 119.3. It was observed that the mass flow rates through individual microchannels in the device vary by up to 45%, depending on the inlet/outlet locations and pressure drop in the manifolds. The results demonstrate that full device simulations provide valuable insight into the multiple parameters that affect cooling performance.

Commentary by Dr. Valentin Fuster
2015;():V001T09A058. doi:10.1115/IPACK2015-48685.

Thermal analysis of electronic devices is essential for designing thermal management systems and for assuring a perfect working condition. In order to have a precise thermal analysis, thermal spreading resistance should be calculated. In this paper, a numerical study is conducted on the thermal resistance of a 2D flux channel with a non-uniform convection coefficient in the heat sink plane. For this purpose, the Finite Volume Method (FVM) is used. As a case study, a 2D flux channel with a discrete specified heat flux and convection edges is assumed. Also, the heat transfer coefficient in the sink boundary condition is determined symmetrically using a hyperellipse function. This function can model a wide variety of different distributions of a heat transfer coefficient from a uniform cooling to the most intense cooling in the central region. All results are compared and validated with the COMSOL commercial software package. The proposed method is useful for thermal engineers for modeling different flux channels with different properties and boundary conditions such as the variable heat transfer coefficient.

Commentary by Dr. Valentin Fuster
2015;():V001T09A059. doi:10.1115/IPACK2015-48698.

The room temperature liquid metal cooling is quickly emerging as a powerful way for the thermal management in many advanced high heat flux devices, spanning from electronics, optoelectronics, battery, to power system etc. Except for its pretty high conductivity that a metal coolant could offer, the unique merit lying behind this new generation cooling strategy is its drivability of the highly conductive coolant through the electromagnetic effect where no moving elements are involved and thus only very few energy consumption is needed. In addition, even waste heat could be strong enough to generate applicable electricity for such flow driving purpose. More directly, the temperature gradient intrinsically generated between the heat source and the sink has also been managed to drive the flow of the coolant and realize an automatic practical enough cooling in some situations. All these practices lead to a totally noiseless pumping of the heat delivery and a compact and reliable cooling modular can thus be possible. Starting from this basic point, we are dedicated here to present an overview on the art and science in developing the technical strategies for a smart driving of the liquid metal cooling of the target devices. Designing philosophy for an innovated thermal management will be discussed. Particularly, electromagnetic pumping, waste thermoelectricity driving, thermosyphon flow effect, etc. will be comparatively evaluated with each of the working performances interpreted. Power consumption rate and efficiency will be quantitatively digested. Typical application examples in the cooling of a series of device areas will be illustrated. Further improvement on the cooling solution along this category will be suggested. Challenging issues in pushing the new technology into large scale utilization will be raised. It is expected that such silent self-driving of the liquid metal coolant will find unique and important values in a wide variety of thermal management areas where reliability, compactness, low noise and energy saving are urgently requested.

Commentary by Dr. Valentin Fuster
2015;():V001T09A060. doi:10.1115/IPACK2015-48770.

The increasing addition of hot components on a single module comes with the challenge to cool the module. The cooling challenges are due to the presence of hot spots and the reduce space for attaching heat sinks to each component. The proposed cooling solution for a multi-chip module is a single vapor chamber shared amongst the multiple heat sources. Designing an application-optimized vapor chamber requires detailed understanding of the different processes occurring, including heat transfer by conduction, two-phase heat transfer, and fluid mechanics in porous media. In this work, a relatively large module with many heat sources is considered. We present a case of two 100W heat sources surrounded by eight heat sources of 10W each. We explore different configurations and their capillary limitations for a vapor chamber (110 mm × 110 mm). We present the comparison of numerical results using two point flux approximation method, CFD model, and a simplified model using potential flow theory to represent flow in the porous media. Results are used to analyze the capillarity limitation of the large vapor chamber in delivering liquid flow to heat source locations for steady state.

Commentary by Dr. Valentin Fuster
2015;():V001T09A061. doi:10.1115/IPACK2015-48802.

This paper describes the effect of variation of energy relaxation time on temperature distribution of power Si MOSFET in electro-thermal analysis. In previous our studies, thermal properties of power Si MOSFET are evaluated using electro-thermal analysis. However, in our previous calculation, energy relaxation time has been assumed to be constant at 0.3 ps, which is widely used value in electro-thermal analysis. This is because energy relaxation time cannot be calculated by classical physics, and it is difficult to detect exact energy relaxation time. However, energy relaxation time is important for evaluating heat generation in electro-thermal analysis. One method to obtain energy relaxation time is Monte Carlo simulation. In this research, we performed Monte-Carlo simulation, and electrical field and lattice temperature dependencies of energy relaxation time were evaluated. Then, we performed electro-thermal analysis of power Si MOSFET with various energy relaxation times, and the effect of change of energy relaxation time on temperature distribution of power Si MOSFET in electro-thermal analysis was discussed. Energy relaxation time in the range of 0.1–1000 kV/cm of electrical field was evaluated in Monte Carlo simulation. The results of Monte-Carlo simulation showed that maximum energy relaxation time becomes about 0.6 ps, and minimum energy relaxation time is about 0.30 ps. Following the results, to investigate the effect of variation of energy relaxation time on temperature distribution of power Si MOSFET, we changed energy relaxation time in electro-thermal analysis, and thermal properties of power Si MOSFET was calculated. The results of electro-thermal analysis showed that energy relaxation time has an effect on temperature distribution of power Si MOSFET. Therefore, accurate energy relaxation time should be considered in electro-thermal analysis for appropriate temperature distribution of power Si MOSFET.

Commentary by Dr. Valentin Fuster

Thermal Management: Thermal Management: Phase Change Materials

2015;():V001T09A062. doi:10.1115/IPACK2015-48082.

The use of nanoparticles to improve the thermal properties of low thermal conductivity phase change materials is of significant interest. However, the addition of nanoparticles to a base fluid is known to result in an increase in viscosity. An increase in viscosity can suppress convective currents, reducing overall heat transfer thus it necessary to quantify the impact of nanoparticle addition on the viscosity of a PCM.

In this work nanoparticle enhanced phase change mateirals are synthesized using paraffin and three different types of nanoparticles: exfoliated graphite nanoplatelets (xGNP), multi-walled carbon nanotubes (MWCNT) and herringbone graphite nanofibers (HGNF). The particles are loaded at rates between 0.0024wt% to 0.1wt%. The viscosity is analyzed at temperatures between 60 and 100°C. The influence of temperature, nanoparticle type and nanoparticle loading level on viscosity are presented and discussed.

The results show that for xGNP and HGNF within the operating condition studied here that there is no impact of the nanoparticle addition on the viscosity of the base material. However, the addition of MWCNT is found to increase the viscosity of the base fluid with the impact increasing with loading level.

Commentary by Dr. Valentin Fuster
2015;():V001T09A063. doi:10.1115/IPACK2015-48369.

The present study deals with transient thermal management using phase change materials (PCMs). These materials can absorb large amounts of heat without significant rise of their temperature during the melting process. This effect is attractive for passive thermal management, particularly where the device is intended to operate in a periodic regime, or where the relatively short stages of high power dissipation are followed by long stand-by periods without a considerable power release. Heat transfer in PCMs, which have low thermal conductivity, can be enhanced by fins that enlarge the heat transfer area. However, when the PCM melts, a layer of liquid is growing at the fins creating an increasing thermal resistance that impedes the process.

The present work aims to demonstrate that performance of a latent-heat thermal management unit may be considerably affected by achieving a so-called close-contact melting (CCM), which occurs when the solid phase is approaching a heated surface, and only a thin liquid layer is separating between the two. Although CCM was extensively studied in the past, its possible role in finned systems has been revealed only recently by our group. In particular, it depends heavily on the specific configuration of the fins.

In the present work, close-contact melting is modeled analytically for a geometry which includes two symmetrically inclined fins. A quasi-steady approach is used for calculating the rate of melting based on the force and energy balances.

The results are expressed in terms of the time-dependent melt fraction and Nusselt number, showing their explicit dependence on the Stefan and Fourier numbers. Moreover, the approach used in the present study may be applied to other geometries in which the heated surface is not horizontal or where there are a number of heated surfaces or fins.

Commentary by Dr. Valentin Fuster
2015;():V001T09A064. doi:10.1115/IPACK2015-48499.

Limited heat dissipation and increasing power consumption in processors has led to a utilization wall. Specifically due to high transistor density, not all processors can be used continuously without exceeding safe operating temperatures. This is more significant in mobile electronic devices which, despite relatively large chip area, are limited by poor heat dissipation — primarily natural convection from the exposed surfaces. In the past, solid-to-liquid phase change materials (PCMs) have been employed for passive thermal control — absorbing energy during the phase change process while maintaining a relatively fixed temperature. However, the lower thermal conductivity of the liquid phase after melting often limits the heat dissipation from the PCM, and in the liquid state, the material can flow away from the desired location. Here we focus on characterization of thermal performance of PCMs with the goal of evaluating dry (gel-to-solid/amorphous-to-crystalline) phase change materials which are intended to mitigate the pumpout issue. Critical thermophysical properties include the thermal conductivity, heat capacity, and latent heat of the phase/state change. The thermal resistance throughout the phase change process is measured by in-house rig which miniaturizes the reference bar method for use with infrared temperature sensing.

Commentary by Dr. Valentin Fuster
2015;():V001T09A065. doi:10.1115/IPACK2015-48522.

Organic phase change materials (PCMs) such as paraffins or unsaturated acids use the latent heat of melting for thermal energy storage as a passive cooling mechanism for portable electronics. Researchers have suggested that a PCM’s thermal energy storage capability is linked to its thermal properties, yet this connection has not yet been quantified. This study first uses group theory and known values from literature to obtain the thermophysical properties for a variety of paraffins and unsaturated acids. Then, multiphysics-based finite element analysis (FEA) is applied to determine the influence of these thermophysical properties on the PCM latent heat storage capability for a side heating configuration. The FEA models include melting and re-solidification, natural convection, conduction, and the monitoring of input and output periodic heat fluxes. The phase change was achieved through application of temperature-dependent viscosity and heat capacity relations. The thermal energy storage efficiency is defined as one minus the ratio of integrated output heat flux to the integrated input heat flux. The FEA results are used to provide predictions of thermal energy storage for a variety of PCMs for various aspect ratios under different heating conditions. Insights are gained in relating thermal storage efficiency to the system configuration.

Commentary by Dr. Valentin Fuster

Thermal Management: Thermal Management: Stacked Die and Multi-Chip-Module and Packaging

2015;():V001T09A066. doi:10.1115/IPACK2015-48042.

Three-dimensional (3D) packaging technology is directly related to the increasing I/O number as stacking chips. This technology has the potential to produce integrated circuits with a much better combination of cost, functionality, performance and power consumption. However, stacked chips raise several thermal issues that need to be addressed and eliminated. In this study, a quantitative study of the conventional solder-based interconnection is conducted based on many different cases of thermal loading, using finite element analysis (FEA). This preliminary study clearly shows limitation of the solder-based interconnection in the thermal management perspective. Underfill for microbμmp acts as a barrier of heat transfer in the conventional 3D stacked chip packages. Therefore, as an alternative, Cu-to-Cu direct bonding (CuDB), which has a better thermal conductivity, is proposed. Its parametric study is performed under the same/different loading conditions and dimensions. This study helps to highlight the thermal behavior of 3D packages consisting of various interconnections. Finally, based on the results, we can propose qualitative design guidelines of 3D packaging depending on various environment and conditions.

Commentary by Dr. Valentin Fuster
2015;():V001T09A067. doi:10.1115/IPACK2015-48104.

Accurate estimation of the thermal conductivity of logic-memory and memory-memory interfaces, between stacked die in 3D microelectronic packages, is key to effective design and early estimates of performance and reliability. Typically, interconnect layers contain hundreds to a few thousands of bumps. Hence lumped/compact modeling of this interfacial layer is essential to reduce computational time and complexity. The typical approach to this lumped modeling is to estimate the effective conductivity of the layer by assuming the bumps and underfill regions can be modelled as parallel thermal resistances (referred to as the volumetric method). This work demonstrates that the volumetric method can significantly underpredict 3D stack thermal resistance and junction temperatures. An alternative method-referred to as the single bump method-of estimation of the thermal conductivity of interconnect layers in 3D stacked-die packages is presented. Studies demonstrate that the proposed single bump method captures the heat transfer in these interfaces accurately. Validation of the single bump modeling is presented by comparing the single bump and volumetric methods with fully discretized models. This comparison also demonstrates that the prevalent volumetric method overestimates the effective thermal conductivity of the interface, while the single bump approach results in more accurate assessment of 3D stack resistance.

Topics: Modeling
Commentary by Dr. Valentin Fuster
2015;():V001T09A068. doi:10.1115/IPACK2015-48316.

In this paper, we present the experimental characterization of 3D packages using a dedicated stackable test chip. An advanced CMOS test chip with programmable power distribution has been designed, fabricated, stacked and packaged in molded and bare die 3D packages. The packages have been experimentally characterized in test sockets with and without cooling, and soldered to the PCB. Using uniform and localized hot spot power distribution, the thermal self-heating and thermal coupling resistance and the lateral spreading in the 3D packages have been studied. Furthermore, the measurements have been used to characterize the thermal properties of the epoxy mold compound and the die-die interface and to calibrate a thermal model for the calculation of equivalent properties of underfilled μbump arrays. This model has been applied to study the trade-off between the stand-off height reduction and the underfill thermal conductivity increase in order to reduce the inter die thermal resistance.

Commentary by Dr. Valentin Fuster
2015;():V001T09A069. doi:10.1115/IPACK2015-48353.

An overview of the thermal management landscape with focus on heat dissipation from 3D chip stacks is provided in this study. Evolutionary and revolutionary topologies, such as single-side, dual-side and, finally, volumetric heat removal, are benchmarked with respect to a high-performance three-tier chip stack with an aggregate power dissipation of 672 W. The thermal budget of 50 K can be maintained by three topologies, namely, 1) dual-side cooling, implemented by a thermally active interposer, 2) interlayer cooling with 4-port fluid delivery and drainage at 100 kPa pressure drop, and 3) a hybrid approach combining interlayer with embedded back-side cooling.

Of all the heat-removal concepts, interlayer cooling is the only approach that scales with the number of dies in the chip stack and hence, enables extreme 3D integration. However, the required size of the microchannels competes with the requirement of low TSV heights and pitches. A scaling study was performed to derive the TSV pitch that is compatible with cooling channels to dissipate 150 W/cm2 per tier. An active IC area of 4 cm2 was considered, which had to be implemented on the varying tier count in the stack. A cuboid form factor of 2 mm × 4 mm × 2.55 mm results from a die count of 50. The resulting microchannels of 2 mm length allow small hydraulic diameters and thus a very high TSV density of 1837 1/mm2. The accumulated heat flux and the volumetric power dissipation are as high as 7.5 kW/cm2 and 29kW/cm3, respectively.

Commentary by Dr. Valentin Fuster
2015;():V001T09A070. doi:10.1115/IPACK2015-48422.

As thermal management techniques for 3D chip stacks and other high power density electronic packages continue to evolve, interest in the thermal pathways across substrates containing a multitude of conductive vias has increased. To facilitate the use of numerical models that can reduce computational costs and time in the thermal analysis of through-layer via (TXV) structures, much research to date has focused on defining effective anisotropic thermal properties for a pseudo-homogeneous TXV medium using isothermal boundary conditions. While such an approach eliminates the need to model heat flow through individual vias, the resulting properties can be shown to depend on the specific boundary conditions applied to a unit TXV cell. More specifically, effective properties based on isothermal boundary conditions fail to capture the local “micro-spreading” resistance associated with more realistic heat flux distributions and local hot spots on the surface of these substrates.

This work assesses how the thermal spreading resistance present in arrays of vias in interposers, substrates, and other package components can be properly incorporated into the modeling of these arrays. We present the conditions under which spreading resistance plays a major role in determining the thermal characteristics of a via array and propose methods by which designers can both account for the effects of spreading resistance and mitigate its contribution to the overall thermal behavior of such substrate-via systems. Finite element modeling of TXV unit cells is performed using commercial simulation software (ANSYS).

Compactly stated, micro-spreading contributes to the total resistance RT = R1d + (fu + fl)Rsp,max, where 0≤ f ≤ 1 are adjustment factors that depend on the conditions at the upper and lower surfaces of the via array layer and Rsp,max occurs under worst-case conditions.

Topics: Modeling
Commentary by Dr. Valentin Fuster

Thermal Management: Thermal Management: Thermal Interface Materials and Heat Spreading

2015;():V001T09A071. doi:10.1115/IPACK2015-48083.

This paper focuses on developing a reliable thermal interface material (TIM) using low melt alloys (LMAs) containing gallium (Ga), indium (In), bismuth (Bi), and tin (Sn). The investigation described herein involved the in situ thermal performance of the LMAs as well as performance evaluation after accelerated life cycle testing, which included isothermal aging at 130°C and thermal cycling from −40°C to 80°C. Three alloys (75.5Ga &24.5In, 100Ga, and 51In, 32.5Bi &16.5Sn) were chosen for testing the thermal performance. Testing methodologies used follow ASTM D5470 protocols and the results are compared with some commercially available TIMs. The LMAs-substrate interaction was investigated by applying the alloys using different surface treatments (copper and tungsten). Measurements show that the alloys did survive extended aging and cycling depending upon the substrate-alloy combinations.

Topics: Alloys , Durability
Commentary by Dr. Valentin Fuster
2015;():V001T09A072. doi:10.1115/IPACK2015-48146.

With increasingly high powers on processors, memories, and chipsets, the voltage regulators (VR) become heavily loaded and a heatsink is often required to prevent overheating the surrounding components on the board. For VR heatsink designs, thermal interface silicone gap filler pads are often used and there is an increasing need to improve VR thermal solutions by reducing thermal resistance of the TIM. A series of TIM2 thickness and performance measurements based on thermal testing was performed in order to understand gap filler characteristics, optimize TIM performance, and utilize best retention design. By utilizing a VR thermal and mechanical test board in wind tunnel testing using the same VR heatsink, thermal performance of TIM2 using gap filler pads over a range of airflow velocities can be measured and compared. The study shows how the optimum TIM performance can be achieved by using the gap filler pads with appropriate thickness for the given designed heatsink standoff heights. The benefit of choosing the right thickness pads over others can be significant and is a valuable learning that can be applied to future VR heatsink designs. Furthermore, the silicone gap filler characteristics and its relationship to board bending and result TIM thickness and thermal performance are investigated and further improved. The learnings help understand the limitations and where the area of improvement can be for future VR heatsink designs.

Commentary by Dr. Valentin Fuster
2015;():V001T09A073. doi:10.1115/IPACK2015-48302.

The thermal contact resistance (TCR) is the crucial issue in the field of heat removal from systems like electronic equipment, satellite thermal control systems, and so on. To cope with the problem, a lot of studies have been done mainly for flat rough surfaces. However, as pointed out so far, there are still wide discrepancies among measured and predicted TCRs, even for similar materials. To investigate the key factors for the abovementioned discrepancies, a fundamental analysis was conducted in our previous study [1] using a simple contact surface model, which was composed of the unit cell model proposed by Tachibana [2] and Sanokawa [3]. Furthermore, by introducing a 2-D microscopic surface model, which consists of random numbers and Abbott’s bearing area curve, the effects of surface waviness and roughness on the temperature fields near the contact interface have been investigated microscopically [4]. In this study, based on a 1-D wavy surface model, a fundamental study has been conducted to predict TCR and the thermal contact conductance (TCC), which is a reciprocal of TCR, between wavy surfaces with the thermal interface material (TIM) under a relatively low mean nominal contact pressure of 0.1–1.0 MPa. From comparison between the calculated and measured results, it has been shown that, in spite of a simple 1-D analysis, the present model predicts the temperature drop at the contact interface, which is obtained as the product of TCR and the heat rate flowing through TIM, within some 10 to 60% error for a TIM with the thermal conductivity of 2.3 W/(m·K) and the initial thickness of 0.5, 1 and 2 mm.

Commentary by Dr. Valentin Fuster
2015;():V001T09A074. doi:10.1115/IPACK2015-48303.

Recently, the increase in performance of semiconductor devices has been accompanied by a simultaneous rise in the amount of heat they emit. Thermal interface materials (TIMs) play an important role in connecting different surfaces to ensure efficient transfer of heat [1]. TIMs can be classified into the following three types: Type 1 materials are greases and pastes, Type 2 materials consist of viscoelastic solids such as gap fillers (an example of which is silicone sheeting), and Type 3 materials are non-compressible solids. Typical Type 1 materials consist of liquid and solid fillers which have a uniform diameter [2]. When pressure is applied, they can easily fill the gap between a heat source and a heat sink, and the gap thickness (BLT: Bond Line Thickness) will eventually decrease to the diameter of the filler. It is well known that under these conditions, the total thermal resistance from a heat source to a heat sink can be minimized. In this study, we demonstrate that we can clearly detect whether the TIM reaches this minimum thickness or not, by measuring the electric capacitance with an LCR meter, instead of evaluating heat related parameters. We performed experiments for a Type 1 TIM which has a viscosity of 240 Pa·s. We hope this technology will be effective, especially in cases of mass-production, because electrical measurement is faster, more repeatable, and more accurate than thermal measurement [3].

Topics: Capacitance
Commentary by Dr. Valentin Fuster
2015;():V001T09A075. doi:10.1115/IPACK2015-48535.

We present a study on the effective mechanical compliance of porous aluminum foams. We develop an experimental setup to characterize the elastic properties as well as evaluate surface deformation with respect to porosity as well as pore size in an effort to correlate the properties to contact resistance of the foams when used as thermal interface materials. There have been multiple studies in the past to evaluate the effective elastic modulus of porous structure as a function of porosity through experimentation, simulation as well as analytic models. This work also serves as a validation for analytic and experimental data published by various researchers in the past. This study is one aspect of a larger study to empirically correlate the area of contact to thermal contact resistance. We evaluate samples with three different porosity and three different PPI (pores per inch) specification. Additionally we analyze effect of presence of filler material in the voids — a phase change material. The filler is used as separate stand-alone TIM in the industry currently.

Commentary by Dr. Valentin Fuster
2015;():V001T09A076. doi:10.1115/IPACK2015-48634.

A heat sink/electronic component system using a phase change interstitial material (thermal interface material) can be optimized by using an adaptive clamping system that adjusts to the flow of the phase change material as it sets with a rise in temperature. The particular thermal clamping system being evaluated in this white paper is that of a servo motor inverter. This thermal system consists of a standard heat sink-fan forced air cooling system. This concept could be applied to any system requiring a phase change thermal interface material.

Commentary by Dr. Valentin Fuster
2015;():V001T09A077. doi:10.1115/IPACK2015-48732.

Within the paper the efficiency of heat transfer through the thermal interface made of three different sintered nanosilver based materials was considered. The efficiency of heat conduction was analyzed by using the IR camera to monitor the changes of junction temperature in the chip of power transistor attached by different TIMs to copper lead-frame. Because in each case the transistor was powered in the same way so the lower difference between junction temperature and substrate temperature meant the better heat transfer efficiency. Moreover by using the X-ray inspection the structural analysis of interface was done.

Commentary by Dr. Valentin Fuster
2015;():V001T09A078. doi:10.1115/IPACK2015-48735.

Self catalyzing Fecralloy substrates are investigated as a growth substrate for Multi-walled Carbon Nanotubes (MWNT) Thermal Interface Materials (TIMs). Fecralloy is used without any additional catalyst and with minimal surface preparation to grow double-sided MWNT TIM assemblies. The growth behavior is studied by way of the array morphologies, i.e. array height, density, crystallinity, and diameter distribution. The effects of growth temperature and time are used to observe the growth kinetics, showing a bimodal growth rate with temperature and an optimal growth rate at 725°C with a noticeable onset of amorphous carbon at higher temperatures. The contact resistance of dozens of such samples are evaluated using a DC, 1D reference bar, thermal conductivity measurement system. Temperature and pressure dependent measurements offer insight into the interfacial phonon conduction physics and elastic deformation mechanics of the CNTs tips respectively. Due to the challenges associated with deliberately controlling a single array morphology, a multivariate, statistical approach is used to observe the trends of contact resistance. The contact resistance shows the strongest correlation with array height, following a R ∼= L−0.5, which contradicts Fourier’s law. However, this is likely a result of the mechanical compliance rather than a ballistic conduction mechanism. Finally, several attempts were made at modeling the relationship between the measured array morphologies and the contact resistance. However, the modeling is relatively unsuccessful, forcing one to rely on the empirical relations found in the exploratory data analysis.

Commentary by Dr. Valentin Fuster

Thermal Management: Thermal Management: Thermoelectrics

2015;():V001T09A079. doi:10.1115/IPACK2015-48421.

Modern CPUs generate considerable wasted heat due to increased power dissipation from high-performance computation. Lots of research effort has extensively focused on using thermoelectric generators (TEGs) to harvest CPU waste heat to increase overall system energy efficiency. To harvest waste heat using TEGs requires a significant temperature differential between the processor as a heat source and the heat spreader/heat sink, as well as a high heat flow. However, the heat-to-electricity conversion efficiency is typically limited to 15 to 20 percent, due to large heat conductivity, low Seebeck coefficient, and low figure of merit of TEGs. In addition, TEGs on a CPU could significantly increase CPU junction temperature compared to the baseline CPU temperature due to its high thermal resistance. Contrary to using TEGs to harvest waste heat from a fixed, spatial temperature differential, this paper presents an approach to harvest CPU waste heat using pyroelectric (PE) materials from the time-varying, temporal temperature differential that is common in current processors. PE materials can generate electricity when subjected to a temporal temperature gradient. The operation of PE materials is distinctly different from TEGs and they have the following advantages. First, the theoretical efficiency is up to 50% using thin films. Second, the overall optimization of PE material is easier than thermoelectric material, since the conversion ratio, the ratio of net harvested energy divided by the heat taken from the hot reservoir, of PE material is independent of the material properties, whereas that of TEG is highly dependent on material properties. Although PE material is also a long-researched energy harvesting material, it is less explored by researchers compared to TEG in the application domain of processor waste heat management. In this paper, we review current PE materials in terms of pyroelectric coefficient and thermal conductivity, and also investigate the harvested power generation from CPU waste heat in a modern computing system.

Topics: Waste heat
Commentary by Dr. Valentin Fuster
2015;():V001T09A080. doi:10.1115/IPACK2015-48682.

We present a comprehensive analysis and optimization of the thermoelectric (TE) heat pump and refrigeration in contact with two constant-temperature reservoirs, followed by a discussion of their cost effectiveness. In many applications in electronics cooling, the heat source temperature is constrained as well as the gas or liquid cooling heat sink. We optimize the thermoelectric design by changing both the element (leg) thickness and drive current simultaneously in order to achieve maximum energy efficiency, i.e., to obtain the highest coefficient of performance (COP) for the heat pump. Each variable and performance is considered per unit area.

COP vs cooling capacity, which is the heat amount pumped, by changing the driving current, shows a unique characteristic and it looks like the Greek character ‘beta’ in a plot. This ‘beta plot’ gives a global view of the performance of various TE heat pump systems. We discuss the similarity with the graph obtained in power generation in contact with the constant temperature reservoirs when the trade-off between the efficiency and power output is considered. In this plot, the maximum COP is found at a much smaller current compared to the maximum heat cooling capacity Qmax. This Qmax is found when the internal resistance is sqrt (1 + ZT) times the sum of the external resistances, but only when these contacts are symmetric and the net temperature difference is zero. The ratio increases slightly as the net temperature difference increases (heat pumping to a higher temperature). This shows some differences compared to the power generation mode where an impedance match happens when the ratio of internal to external resistances is constant at sqrt (1 + ZT). If the contact thermal resistances with the hot and cold sides are asymmetric, Qmax and the optimum resistance ratio are both reduced when the heat sink resistance increases and they both increase as the heat sink resistance decreases.

TE materials are expensive relative to the other components; hence, it is important to minimize the material use. The COP per cost and cooling capacity per cost are investigated. Similar to power generators, the TE element can be thinner as the fractional area coverage of the TE elements is reduced, while maintaining a constant internal thermal resistance. The most cost effective design is found to be thinner than that of the maximum performance. Also, the ZT value impact for the cost performances is smaller, especially in COP.

Commentary by Dr. Valentin Fuster
2015;():V001T09A081. doi:10.1115/IPACK2015-48692.

We present an analytic model and optimization of impingement heat transfer in fluid-to-fluid heat exchangers with integrating a thermoelectric (TE) generator between the hot and cold fluid flows. In power generation systems, designing for maximum power output generally involves balancing the external thermal resistances while the generator contacts the hot and cold temperature reservoirs. In fluid-to-fluid heat exchangers, fluid temperatures are not constant or uniform. They gradually change along the flow direction. In general, counter-flow heat exchangers outperform parallel flow configurations in maximizing TE power generation using internal fluid flows. We show here the performance of our impingement model compared with a counter-flow configuration as the base line.

To obtain the maximum power output from practical thermoelectric materials (ZT values are 1.2–1.8), the enhancement of liquid-to-wall heat transfer is significant. An array of traditional impinging jet orifices provides a uniformly planar and focused heat transfer process that spatially targets the TE elements. This approach provides more uniform hot and cold side temperatures among the TE elements. We investigate the impact of introducing impingement orifices directly at the locations of the TE elements. The major focus of this work is the trade-off between the advantage of increasing power generation by impingement and the disadvantage of introducing additional pressure drop.

Decreasing the external thermal resistances yields not only a larger maximum power output but also requires thinner TE elements. This enables lower cost per power generation capacity approaching the 0.2–0.3 $/W range as well as a more lightweight design. We report here the associated cost impacts for the impinging jet arrangement.

Design optimization depends on the specific constraints and parameters, such as TE material and substrate thickness, flow design to avoid the stagnation, and required exit temperatures. In some cases, active pumping by an additional actuator can augment the enhancement, while a fraction of generated power is consumed for the actuation. In the paper, we show examples of gas and liquid flow cases.

Commentary by Dr. Valentin Fuster

Thermal Management: Thermal Modeling Methodologies and Simulations

2015;():V001T09A082. doi:10.1115/IPACK2015-48205.

The thermal network method has a long history with thermal design of electronic equipment. In particular, a one-dimensional thermal network is useful to know the temperature and heat transfer rate along each heat transfer path. It also saves computation time and/or computation resources to obtain target temperature. However, unlike three-dimensional thermal simulation with fine pitch grids and a three-dimensional thermal network with sufficient numbers of nodes, a traditional one-dimensional thermal network cannot predict the temperature of a microprocessor silicon die hot spot with sufficient accuracy in a three-dimensional domain analysis. Therefore, this paper introduces a one-dimensional thermal network with average temperature nodes. Thermal resistance values need to be obtained to calculate target temperature in a thermal network. For this purpose, thermal resistance calculation methodology with simplified boundary conditions, which calculates thermal resistance values from an analytical solution, is also introduced in this paper. The effectiveness of the methodology is explored with a simple model of the microprocessor system. The calculated result by the methodology is compared to a three-dimensional heat conduction simulation result. It is found that the introduced technique matches the three-dimensional heat conduction simulation result well.

Commentary by Dr. Valentin Fuster
2015;():V001T09A083. doi:10.1115/IPACK2015-48277.

In this study, the thermal and hydraulic performances of an air-cooled planar heat sink with cross-connected alternating converging-diverging channels were investigated. The commercial CFD solver ANSYS Fluent was used to solve the fluid flow and heat transfer from the fins to the air flow for a range of converging-diverging channel expansion ratios, heat generation rates and Reynolds numbers. The converging and diverging channel sections create high and low pressure zones, respectively, in the flow domain and this pressure difference induces secondary flows from the converging channel sections to the diverging channel sections through the cross connections. The observed heat transfer enhancement results from two different phenomena: (1) thermal boundary layer disruption and re-initialization at the cross connections and (2) fluid mixing; where the former reduces the convection resistance in the vicinity of the fin walls by reducing the thermal boundary layer thickness, while the latter allows a more uniform temperature build-up of air in the streamwise direction. Despite the fact that the pressure drop penalty increases due to flow restriction compared to the straight channel heat sinks, it is possible to enhance the Nusselt number for up to 100% with the proposed heat sink design.

Commentary by Dr. Valentin Fuster
2015;():V001T09A084. doi:10.1115/IPACK2015-48362.

Multi-chip packages (MCPs) based solutions are becoming increasingly adopted as it results in higher signal count, density and enables increasing bandwidth demands and allows for heterogeneous integration [1,2]. However, manufacturing tolerances impose a variability in these stacks which results in new requirements for thermal interface materials. This paper describes the thermal, mechanical, and reliability challenges associated with MCP packages, and highlights need for novel thermal interface materials.

Commentary by Dr. Valentin Fuster
2015;():V001T09A085. doi:10.1115/IPACK2015-48370.

The information technology (IT) industry is exploring three-dimensional (3D) stacking of chips to maintain future computing scalability. However, 3D chip stacks require a solution to significant new thermal challenges. Interlayer two-phase evaporative cooling with a chip-to-chip interconnect-compatible dielectric fluid is an enabling technology but faces significant development issues. One such issue is the inability to thermally model a microprocessor with spatially varying heat sources together with a two phase microfluidic convection network. While progress has been made on two-phase conjugate simulations at the chip and channel levels, none of those provide a computationally manageable approach.

In the present study, a reduced physics conjugate heat transfer model has been developed for simulating two-phase flow boiling through chip embedded micron scale cavities. This model has been validated with good accuracy against data available from literature. The validated model was then extended to predict the thermal performance of a state-of-the-art microprocessor chip with embedded two-phase cooling, where significant improvements in device junction temperatures were observed compared to the baseline cooling solution.

Commentary by Dr. Valentin Fuster
2015;():V001T09A086. doi:10.1115/IPACK2015-48568.

Continued CMOS scaling accompanied with a stall in the voltage scaling has led to high on-chip power densities. High on-chip power densities elevate the temperatures, substantially limiting the performance and reliability of computing systems. The use of Phase Change Materials (PCMs)1 has been explored as a passive cooling method to manage excessive chip temperatures. The thermal properties of PCMs allow a large amount of heat to be stored at near-constant temperature during the phase transition. This heat storage capability of PCM can be leveraged during periods of intense computation. For systems with PCM, development of new management strategies is essential to maximize the benefits of PCM. In order to design and evaluate these management strategies, it is necessary to have an accurate PCM thermal model. In our recent work, we proposed a detailed phase change thermal model, which we integrated into a compact thermal simulation tool, HotSpot. In this paper, we build a hardware testbed incorporating a PCM unit on top of the chip package. We then validate the accuracy of our previously proposed thermal model by comparing the HotSpot simulation results against the measurements on the testbed. We observe that the error between the measured and simulated temperatures is less than 4°C with 0.65 probability. Finally, we implement a soft PCM capacity sensor that monitors the remaining PCM latent heat capacity to be used for development of thermal management policies. We evaluate a set of thermal management policies on the testbed. We compare policies that adjust the sprinting frequency based on current temperature against the policies that take action based on the remaining PCM capacity.

Topics: Hardware
Commentary by Dr. Valentin Fuster
2015;():V001T09A087. doi:10.1115/IPACK2015-48606.

Thermal modeling of concealed Remote Radio Head (RRH) units which are placed on a cell tower is the main interest of this paper. RRH units which dissipate fixed amount of heat are modeled in higher ambient temperature. The effect of creating an enclosure on the open air assembly of RRH units is studied. The aim of this study is to analyze if concealment of the RRH units is possible under natural convection for cooling under boundary conditions with inlet temperature 55°C, solar loading and wind. Also, the study has been made to investigate if stacking of one, two, or three concealed sections is possible under natural convection cooling process.

Various configurations were modeled and analyzed using a computational fluid dynamics (CFD) tool. Different temperature profiles are reported. An enclosure has been created and the operating temperatures of the enclosed model and an open model are compared. After studying the air flow pattern in the model, specific design modifications, like placing baffles on top of each RRH unit, are suggested for proper air flow management thereby facilitating efficient thermal management inside the enclosure. Also, addition of baffles helped in removing the hot spots in the model.

Stacking concealed sections poses a challenge since hot exhaust air from a lower section may enter the air inlets of top sections. Thereby, higher surface temperatures on RRH units in the upper sections are observed. Orientation of the upper section has been changed to address this issue. Results show that with proper airflow management and arrangement of sections, it is possible to have three sections one on top of the other.

Topics: Modeling
Commentary by Dr. Valentin Fuster
2015;():V001T09A088. doi:10.1115/IPACK2015-48678.

Presented are the results of a 3-D numerical analysis of a composite heat spreader for immersion cooling of a 20 × 20 mm microprocessor. The spreader is comprised of two 0.5 mm thick Copper (Cu) laments separated by a layer of highly ordered pyrolytic graphite (HOPG), 0.25–1.0 mm thick. The exposed surface of the top Cu lament has an average roughness, Ra = 1.79 μm and is cooled by saturation nucleate pool boiling of PF-5060 dielectric liquid. Investigate is the impact of δHOPG on the total power removed, the maximum temperature of the underlying chip, Tmax, and mitigating the chip hot spots. Increasing δHOPG increases the total power removed, but also increases Tmax. The spreader with a 1.0 mm-thick δHOPG is capable of removing 318 W, without exceeding 90% of the critical heat flux (CHF), at Tmax = 120°C. This power removal is significantly higher than that with an all Cu spreader of the same thickness of 90 W, but at much lower Tmax of 67°C. Composite spreaders with δHOPG = 0.25, 0.5, and 0.75 mm are capable of removing up to 160 W at Tmax = 85°C, 228 W at 100°C, and 292W at 115°C, respectively. The HOPG suppresses the transmission of hot spots to the spreader surface and increasing δHOPG does not mitigate the hot spots.

Commentary by Dr. Valentin Fuster
2015;():V001T09A089. doi:10.1115/IPACK2015-48740.

Augmenting the thermal conductivity of polymer materials is actively being attempted by adding one or more fillers with higher thermal conductivity into matrix materials. In this study, the effective thermal conductivity of composite materials was investigated numerically under the effects of the thermal conductivity ratio between two particle fillers and the matrix material, and the particle volume fractions. The results indicate that the effective thermal conductivity of composites containing hybrid filler is higher than that of single filler. The effective thermal conductivity increases with the increase of thermal conductivity ratio between two fillers in general when this ratio is less than unity, and the maximum effective thermal conductivity approaches when this ratio is less than and close to unity. However, this trend is changed when this ratio is greater than unity. Based on the results, a generalized correlation is proposed as a function of four non-dimensional parameters. The results obtained in this study can be widely utilized for predicting the thermal conductivity of hybrid-filler-nanoparticle composite materials.

Commentary by Dr. Valentin Fuster

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In