Data Center Cooling Research Articles

Analysis of Thermal Performance Metrics—Application to CPU Cooling in HPC Servers

Optimization of heatsinks has a critical importance for the energy-efficient cooling of high-performance computing (HPC) servers to meet reliability, performance, and cost targets. This article provides a detailed application of the effectiveness as a performance metric to provide a qualitative comparison of the CPU heatsink solutions. First, the use of effectiveness as a metric to provide an absolute performance of heatsinks on a simple scale of 0–1 is highlighted. Then, the effectiveness and thermal resistances are compared to explore the limits of air cooling and the relative merits of liquid cooling. Second, a response surface method (RSM) for the heatsink optimization is presented. Furthermore, a response surface optimization (RSO) based on the design of experiments (DOE) is performed on the numerical models of the heatsink and cold plates using FloTHERM computational fluid dynamics (CFD) software. The simulation results obtained at 45 °C inlet are used to provide an objective comparison among the optimum heatsink and cold plates. The simulations show that heatsinks have the effectiveness of 0.65–0.75 and a cooling capability of 150–250 W. The cold plates have an effectiveness of 0.25–0.30 and a cooling capability of 325–425 W. Furthermore, the simulation results are compared with benchmark validation studies. Counterintuitively, the higher effectiveness obtained for the air-cooled heatsinks indicates limited scope for further improvement, whereas the cold plates (with lower effectiveness) provide great scope for further improvements to meet future cooling challenges.

Synergetic modulation of power factor and thermal conductivity for Cu3SbSe4-based system

Many methods have been focused on improving the thermoelectric performance of Cu3SbSe4 based materials. However, no effective way can simultaneously achieve the significant improvement of power factor and the remarkable reduction of thermal conductivity. In this work, we introduce AgSb0.98Sn0.02Se2 particles into Cu3Sb0.96Sn0.04Se4 matrix forming Cu3Sb0.96Sn0.04Se4/x wt%AgSb0.98Sn0.02Se2 (x = 0, 1, 3, 5, 7) nanocomposites by milling and spark plasma sintering technique. Due to the reduced thermal conductivity and enhanced carrier mobility, a record-high ZT value of 1.17 is obtained for the sample Cu3Sb0.96Sn0.04Se4/5wt%AgSb0.98Sn0.02Se2. The reduced thermal conductivity is attributed to the enhanced phonon scattering caused by the second phase AgSb0.98Sn0.02Se2 with multiscale size and the enhanced power factor originated from increased carrier mobility due to the reduced carrier concentration, which is caused by the new acceptor energy level related to the interfacial defects. Present results demonstrate that excellent thermoelectric performance can be realized in Cu3Sb0.96Sn0.04Se4 matrix with AgSb0.98Sn0.02Se2 particles embedded.

Near-field radiative heat transfer between twisted nanoparticle gratings

We study the near-field radiative heat transfer between two twisted finite-size polar dielectric nanoparticle gratings. Different from previous studies of the same configuration, we do not rely on any approximated effective medium theory to describe the gratings. By the full many-body radiative heat transfer theory, we are able to investigate how the size, distance, and relative orientation between the gratings influence the radiative heat flux. By changing the twisting angle θ, we show a significant oscillation of the thermal conductance G(θ), due to the size effect for gratings of both square and circular shapes. The distance- and twisting-dependent coupling between the gratings accounts for a strong and characteristic modulation of radiative thermal conductance with implications for the energy management, sensing, and the micro-electromechanical system (MEMS) and nano-electromechanical system (NEMS) devices.

Experimental and numerical heat transfer analysis of heat-pipe-based CPU coolers and performance optimization methodology

This paper presents an experimental and numerical heat transfer analysis of heat-pipe-based CPU coolers and a performance optimization methodology. The first part of the study focuses on the performance of two commercial HP-based CPU coolers under inclination angles of 0°, 90° and 180°. The results show that the 90° orientation provides the best thermal performance. The influence of heat pipe orientation on the performance of the entire system is obscured due to the much higher thermal resistance on the air-side of the cooler. A fourfold increase in air volumetric flow rate has only a minor effect on the cooling performance enhancement with a reduction of the thermal resistance from 0.11 K W−1 to 0.074 K W−1 at the highest heating power. In the second part of the study, heat transfer numerical simulations of the finned part of a cooler were performed and validated using experimental results. The output of the simulation is a 2-D temperature field, which is used as an input for the optimization methodology based on fin effectiveness and fin efficiency. Optimizing the fin geometry by removing unnecessary material yielded a 23% increase of the fin efficiency and decreased the weight of a fin by approx. 30%, proving the usefulness of the proposed methodology, which helps reduce costs, weight and development time of finned HP-based coolers.

Synergistic modulation of power factor and thermal conductivity in Cu3SbSe4 towards high thermoelectric performance

Abstract In this study, we develop a synergistic modulation of the thermal conductivity and power factor of Cu3SbSe4-based materials through Sn and Zr or Hf co-doping by using a facile microwave-assisted solvothermal method. A series of Cu3Sb1-xMxSe4 (M = Zr or Hf, x = 0, 0.02, 0.04, 0.06 and 0.08) compounds are firstly synthesized through the microwave-assisted solvothermal method combined with spark plasma sintering (SPS) process. The effect of Zr and Hf doping on the thermoelectric properties of Cu3Sb1-xMxSe4 (M = Zr or Hf) has been investigated. With increasing the content of Hf and Zr, the thermal conductivity of Cu3Sb1-xMxSe4 (M = Zr or Hf) is obviously decreased to 0.518 Wm−1K−1 of Cu3Sb0.92Hf0.08Se4 and 0.433 Wm−1K−1 of Cu3Sb0.92Zr0.08Se4 at 623 K, respectively. In addition, Sn doping further improves the low electrical conductivity and boosts the power factor, yielding a peak zT value of ~0.82 of Cu3Sb0.91Sn0.03Hf0.06Se4 at 623 K, which is ~228% higher than that of the pristine Cu3SbSe4. Our work provides a new methodology for the decoupling of thermal and electrical properties of the Cu3SbSe4-based thermoelectric materials.

Simulation-Based Assessment of Performance Indicator for Data Center Cooling Optimization

Abstract Data center power consumption has grown substantially in the past 20 years. According to the United States Data Center Energy Usage Report, the data center consumption in 2014 was estimated at 70 billion kWh, which accounted for 1.8% of the total U.S. electricity. The present effort investigates the effects of various data center parameters on a new set of metrics called the performance indicator to help assess and optimize the cooling performance of data centers. The three metrics in the performance indicator include power usage effectiveness ratio (PUEr), thermal conformance, and thermal resilience. The data center parameters investigated include computer room air handler (CRAH) setpoints, room configuration layouts, and containment strategies. The results show that the CRAH setpoint significantly influences the PUEr with higher setpoint values resulting in lower PUEr values. Room configuration layout changes and containment strategies showed substantial effects on thermal conformance and thermal resilience. The thermal conformance was increased approximately 10% with room configuration changes without changing the PUEr. Full hot aisle containment also improved the thermal conformance by 7.4%.

MHD enhanced nanofluid mediated heat transfer in porous metal for CPU cooling

In this study, impingement cooling of a porous metal CPU cooler saturated with nanofluid under the effects of magnetic field was modeled analytically. Darcy-Brinkman-Forchheimer model and viscous dissipation effect were considered to simulate fluid flow of nanofluid through the porous media under magnetic field. Proper similarity variables were proposed and the original partial differential governing equations were converted to nonlinear ordinary differential equations (ODEs), and the resulted ODEs were solved numerically. A system of water-alumina nanofluid flowing through a rectangular porous metal foam with top impinging jet (fan) and hot bottom wall (CPU surface) was modeled. The analytical solver was first validated against CFD and experimental data. Next, effects of different critical parameters including Darcy number (5×10-4≤DaL≤5×10-2), Reynolds number (50≤ReL≤500), aspect ratio (0.25≤H/L≤1.0), Eckert number (0≤EcL≤5×10-2), Hartmann number (0≤HaL≤200), porosity (0.85≤ε≤0.95), and dimensionless distance from the axis of symmetry plane (0≤XL≤1.0), on fluid flow and heat transfer behaviors were investigated. Results indicate that the increase of DaL can enhance heat transfer performance, while opposite trends are found for aspect ratio and Eckert number. However, the behavior of HaL is more complex. At a low porosity, Nusselt number slightly decreases as HaL increases, while an opposite behavior is observed at a high porosity. Overall, the use of a stronger magnetic field is beneficial to enhance the impingement cooling heat transfer with highly porous metal foam.

Efficient thermal conductivity modulation by manipulating interlayer interactions: A comparative study of bilayer graphene and graphite

The application of graphene in high-performance thermal management has received a lot of attention in recent years, which still needs further exploration and development. Here, based on first-principles calculations, the thermal transport is found to be efficiently modulated by enhancing interlayer interactions in bilayer graphene (BLG), showing a different trend compared to graphite. The results of our work suggest that, by enhancing the interlayer force, the “in-plane” anharmonic phonon transport of BLG while the “out-of-plane” harmonic phonon transport of graphite can be effectively tuned. By manipulating interlayer interactions, a controllable and directed parameter (6% out-of-plane compressing deformation of BLG can achieve more than 25% decrement of in-plane thermal conductivity; 10% out-of-plane compressing deformation of graphite can increase out-of-plane thermal conductivity by more than 5 times) for tuning the thermal conductivity can be achieved. The difference in the effect of the interlayer force on thermal conductivity for low-dimensional and bulk materials emphasizes the significance of the anharmonic phonon transport properties of low-dimensional materials with interlayer interaction and thereby provides an important insight for promoting the future application of bilayer graphene and graphite.

Thermal Performance Evaluation of a Data Center Cooling System under Fault Conditions

If a data center experiences a system outage or fault conditions, it becomes difficult to provide a stable and continuous information technology (IT) service. Therefore, it is critical to design and implement a backup system so that stability can be maintained even in emergency (unforeseen) situations. In this study, an actual 20 MW data center project was analyzed to evaluate the thermal performance of an IT server room during a cooling system outage under six fault conditions. In addition, a method of organizing and systematically managing operational stability and energy efficiency verification was identified for data center construction in accordance with the commissioning process. Up to a chilled water supply temperature of 17 °C and a computer room air handling unit air supply temperature of 24 °C, the temperature of the air flowing into the IT server room fell into the allowable range specified by the American Society of Heating, Refrigerating, and Air-Conditioning Engineers standard (18–27 °C). It was possible to perform allowable operations for approximately 320 s after cooling system outage. Starting at a chilled water supply temperature of 18 °C and an air supply temperature of 25 °C, a rapid temperature increase occurred, which is a serious cause of IT equipment failure. Due to the use of cold aisle containment and designs with relatively high chilled water and air supply temperatures, there is a high possibility that a rapid temperature increase inside an IT server room will occur during a cooling system outage. Thus, the backup system must be activated within 300 s. It is essential to understand the operational characteristics of data centers and design optimal cooling systems to ensure the reliability of high-density data centers. In particular, it is necessary to consider these physical results and to perform an integrated review of the time required for emergency cooling equipment to operate as well as the backup system availability time.

A Calculation Model for Typical Data Center Cooling System

This paper has established a typical calculation model for typical data center cooling system, including the inside room CRAC system, outside free cooling usage system and chiller. The aim of establishing such method is to calculate the power consumption of the data center and to guide the work of energy conservation. The results show that, higher inside room temperature, lower water mass flow rate and lower outside air temperature lead to lower PUE.

Experimental study and analytical modeling of thermosyphon loop for cooling data center racks

This paper presents experimental study and analytical model for a passive two-phase thermosyphon loop designed to remove directly the hot air flow inside a data center. The cooling loop with separated lines working without electricity is tested under different conditions. A mini-channel parallel-flow evaporator and a finned-tubes condenser are used to transfer heat from the data center to outside environment. Experimental tests are conducted with different n-pentane fill ratios. According to these tests, the cooling performance is showed and the refrigerant distribution is specified. The proposed steady-state model is based on the combination of the thermal and hydraulic models of the two-phase flow inside the loop. It well predicts the experimental results such as the mass flow rate and working fluid pressure drops. As investigated by the model, the most parameters influencing the cooling capacity are: the working fluid, the condenser tubes number and the outside temperature.

Comparison of Experimental Measurements of Thermal Conductivity of Fe2O3 Nanofluids Against Standard Theoretical Models and Artificial Neural Network Approach

In the present work, the practicability of Fe2O3 nanofluids for heat transfer applications has been examined. Nanofluids performance, in terms of modulation of thermal conductivity, has been investigated with increasing concentration of Fe2O3 nanoparticles in water and ethylene glycol base fluids at 10, 20, 30, 40, 50, 60 and 70 °C. Fe2O3 nanoparticles have been synthesized using the wet chemical method and characterized using TEM, SEM, XRD and UV–Vis. The characterization results revealed a face-centered cubic structure having alpha phase and particle size in the range of 40-55 nm for the synthesized Fe2O3 nanoparticles. Thermal conductivity measurement results show increases in thermal conductivity with the increase in concentration and temperature of nanofluids. 16.45 and 19.76% enhancement in thermal conductivity have been observed for Fe2O3–water and Fe2O3–ethylene glycol nanofluids of 2 vol.% at 70 °C compared to water and ethylene glycol base fluids at 10 °C, respectively. Results of the ANN approach are in good agreement with experimental results, and H–C model gives better predictions compared to other standard models. The study gives clear insights into improved heat transfer performance by material engineering.

Performance Investigation on the Thermosyphon with Evaporative Condenser for Free Cooling of Data Centers

The microchannel heat exchanger is applied as the heat exchanger of the thermosyphon for data center cooling, and the evaporative cooling experiment is carried out on the side of the microchannel condenser. The influence of evaporative cooling on the microchannel condenser is measured by using the enthalpy different laboratory in the experiment, and the temperature difference between the inlet and outlet air of the microchannel condenser and with different indoor and outdoor temperature and humidity was analyzed. The experimental results show that evaporative cooling can make the condenser inlet air temperature achieve a large temperature drop and increase the temperature drop of the inlet and outlet air, while the spraying can cool the air through the thermosyphon about 0.3-1.9 degrees and increase about 0.1-0.6kW heat transfer of the system. Moreover, the evaporative cooling can improve the heat exchange efficiency of condensers, expand the application area of the loop thermosyphon and extend the annual utilization time.

Study on Thermal and Hydrodynamic Performance of a Triple Fluid Heat Exchanger with Different Passes and Rows

Hybrid cooling system has become a promising thermal management system for data center considering its capability of removing heat from various electronic components systematically. Water cooling and air cooling can be used for cooling the active component and passive component respectively. A novel triple fluid heat exchanger (TFHEX) is designed to dissipate the heat from these fluids into the coolant. This paper discusses thermal and hydrodynamic performance of TFHEX that specially designed for hybrid cooling of data. In present work, the THFEX is designed to dissipate heat from 32kW rack in which the server that dissipates 90% of heat from the active component and 10% of from passive component. Since the water side has much higher load, heat transfer from tube to tube in TFHEX has to be carefully maintained without significant increase in pressure drop. It was found that TFHEX can dissipate heat from different heat source at various load without significant increase in the pressure drop if designed carefully and therefore this heat exchanger can be a feasible solution for hybrid cooling data center. The results critically evaluated the trade-off between the heat transfer capacity of TFHEX and its pressure drop i.e. the pumping power requirement. This can be a reference for data center engineers to select the appropriate configuration for TFHEX for hybrid cooling.

Synergistic modulation of mobility and thermal conductivity in (Bi,Sb)2Te3 towards high thermoelectric performance

Two-step sintering efficiently enhances zT by tuning the microstructure in a wide range from atomic defects to micrometer second phase.

Disparate strain response of the thermal transport properties of bilayer penta-graphene as compared to that of monolayer penta-graphene.

In this study, strain modulation of the lattice thermal conductivity of monolayer and bilayer penta-graphene (PG) at room temperature was investigated using first-principles calculations combined with the phonon Boltzmann transport equation. The thermal conductivities of both the monolayer and the bilayer PG exhibit a robust nonmonotonic up-and-down behavior under strain despite the effect of van der Waals (vdW) interactions, and the thermal conductivities of bilayer PG under strain are significantly reduced by up to 87%. Using phonon-level systematic analysis, the variation of thermal conductivity with the increasing strain was determined by increasing the phonon lifetime in specific phonon modes, and that with the reduction of strain was determined by the decrease of both phonon group velocity and phonon lifetime. Moreover, bilayer PG shows an unexpectedly different response to strain when compared with monolayer PG, and a significantly larger reduction (>60%) in the thermal conductivity of bilayer PG is achieved when the strain reaches 10% because the interlayer interactions enhance the phonon anharmonicity of the phonon modes of ultra-low frequency. Our study shows that bilayer PG will have tremendous opportunities for application in thermal management and two-dimensional nanoscale electronic devices owing to its largely tunable thermal conductivity.

Newton-like phasor extremum seeking control with application to cooling data centers

We design a Newton-like phasor-based Extremum Seeking Control (ESC) aimed at multi-variable nonlinear control systems. We first augment the derivative estimator underlying the phasor approach to capture the Hessian of the plant index. Then we use these estimates to steer the system along a Newton-like direction. We assess the closed-loop stability of the proposed scheme and demonstrate its effectiveness in a data center cooling scenario.

Modulation of thermal conductivity and thermoelectric figure of merit by anharmonic lattice vibration in Sb2Te3 thermoelectrics

Anharmonic lattice vibration in metallic and semiconducting Sb2Te3 is investigated using temperature dependent Raman spectroscopic studies, synchrotron powder diffraction, and heat-capacity measurements. Thermal variation of structural parameters divulges temperature dependence of structural anisotropy in both the samples. It is revealed that semiconducting Sb2Te3, having higher defect density and larger phonon anharmonicity, possesses lower thermal conductivity. As compared to its metallic counterpart, significant increase in thermoelectric figure of merit, ZT is reported for the semiconducting Sb2Te3. Correlation among structural anisotropy, phonon anharmonicity and ZT is established.

Joint Cooling and Server Control in Data Centers: A Cross-Layer Framework for Holistic Energy Minimization

In this paper, we try to address the issue of holistic energy minimization in data centers by designing a cross-layer optimization framework. We formulate a mixed integer nonLinear programming problem named data center holistic energy minimization problem, and develop an efficient JOINT algorithm to solve it. JOINT cuts energy consumption via collaboration of system components across different control layers, and dynamically configures system parameters according to the current workload. To evaluate the performance of the proposed framework, we conduct extensive simulations against synthetic and real-world workload traces. Experimental results suggest that our cross-layer framework is effective in data center holistic energy reduction.

Voltage-Controlled Bistable Thermal Conductivity in Suspended Ferroelectric Thin-Film Membranes.

Ferroelastic domain walls in ferroelectric materials possess two properties that are known to affect phonon transport: a change in crystallographic orientation and a lattice strain. Changing populations and spacing of nanoscale-spaced ferroelastic domain walls lead to the manipulation of phonon-scattering rates, enabling the control of thermal conduction at ambient temperatures. In the present work, lead zirconate titanate (PZT) thin-film membrane structures were fabricated to reduce mechanical clamping to the substrate and enable a subsequent increase in the ferroelastic domain wall mobility. Under application of an electric field, the thermal conductivity of PZT increases abruptly at ∼100 kV/cm by ∼13% owing to a reduction in the number of phonon-scattering domain walls in the thermal conduction path. The thermal conductivity modulation is rapid, repeatable, and discrete, resulting in a bistable state or a "digital" modulation scheme. The modulation of thermal conductivity due to changes in domain wall configuration is supported by polarization-field, mechanical stiffness, and in situ microdiffraction experiments. This work opens a path toward a new means to control phonons and phonon-mediated energy in a digital manner at room temperature using only an electric field.