Thermal Management Of Data Centers Research Articles

Measurement of the Thermal Performance of a Custom-Build Single-Phase Immersion Cooled Server at Various High and Low Temperatures for Prolonged Time

Abstract The next radical change in the thermal management of data centers is to shift from conventional cooling methods like air-cooling to direct liquid cooling to enable high thermal mass and corresponding superior cooling. There has been in the past few years a limited adoption of direct liquid cooling in data centers because of its simplicity and high heat dissipation capacity. Single-phase engineered fluid immersion cooling has several other benefits like better server performance, even temperature profile, and higher rack densities and the ability to cool all components in a server without the need for electrical isolation. The reliability aspect of such cooling technology has not been well addressed in the open literature. This paper presents the performance of a fully single-phase dielectric fluid immersed server over wide temperature ranges in an environmental chamber. The server was placed in an environmental chamber and applied extreme temperatures ranging from −20 °C to 10 °C at 100% relative humidity and from 20 to 55 °C at constant 50% relative humidity for extended durations. This work is a first attempt of measuring the performance of a server and other components like pump including flow rate drop, starting trouble, and other potential issues under extreme climatic conditions for a completely liquid-submerged system. Pumping power consumption is directly proportional to the operating cost of a data center. The experiment was carried out until the core temperature reached the maximum junction temperature. This experiment helps to determine the threshold capacity and the robustness of the server for its applications in extreme climatic conditions.

Comparison of data driven modeling approaches for temperature prediction in data centers

Energy-efficient thermal management of data centers based on dynamic optimization and provisioning of cooling resources requires rapid (nearly real-time) predictions of temperatures within data centers. This work for the first time compares multiple Data-Driven Models (DDMs) to achieve such rapid temperature predictions. DDM typically employs statistical or machine learning-based tools, in combination with physics-based modeling and/or experimental data to predict system behavior. In general, DDM approaches are well-suited to systems that have multiple operational states based on interactions between the many electrical, mechanical and control parameters typical of data centers.This study compares the performance of three different DDM methods, namely Artificial Neural Networks (ANN), Support Vector Regression (SVR), Gaussian Process Regression (GPR) in predicting both steady-state and transient rack inlet air temperature distributions in data centers. Additionally, Proper Orthogonal Decomposition (POD) was considered for transient modeling. The data used for training and analysis were obtained by performing 300 offline numerical simulations with a room-level, experimentally validated computational fluid dynamics/heat transfer (CFD/HT) model.The performance of the four data-driven models was evaluated based on the absolute mean error for interpolation and extrapolation, and the adaptability of the models to changes in physical domain (data center room) configuration. Additionally, the impact of the size of the training data set on prediction accuracy is also compared for the four models.For the steady-state case study, the predictions for ANN, SVR and GPR models are in good agreement with CFD/HT simulations, with the GPR model having the smallest overall average prediction error of 0.6 °C in rack inlet air temperature, corresponding to a relative error of 2.7% with respect to rack inlet temperature measured in °C. It was found that for all the frameworks the prediction error increases when the size of training data set was less than 300 samples. The GPR model had the best accuracy for smaller training data sets compared with the other models, with an average prediction error for rack inlet temperatures <1 °C when trained on only 50 simulations. For the transient case study, the interpolative prediction error for all the models is very low (<0.3 °C); however, the extrapolative prediction errors are much greater, and appear to be directly proportional to the (here, temporal) “distance” from the interrogation point to the input parameter space.

데이터센터 공조·냉각시스템 효율평가 위한 정량적 성능지표 도출 및 모니터링 방법

데이터센터 공조·냉각시스템 효율평가 위한 정량적 성능지표 도출 및 모니터링 방법

A new mathematical model for multi-scale thermal management of data centers using entransy theory

Based on entransy theory, this paper proposes a new mathematical model for multi-scale data center thermal management. A comparison of heat transfer optimizing is made first, using exergy/entropy method and entransy dissipation, to verify the fitness of entransy theory. Based on which, a complete model of entransy dissipation for data center heat transfer is built, from CPU level to data center room level, with detailed computational derivation. Specifically, the calculating method of entransy dissipated by undesired air mixing has been derived, which can give quantitative evaluation on air mixing to the entire thermal performance of data center. A case study of a CFD simulation and a CRAC retrofitting engineering have been performed to verify the entransy model. In the case study, the temperature penalty caused by undesired air mixing is calculated using the entransy analysis model and testified by the retrofitting test, which directly reduces the free cooling potential and decrease data center energy performance. This entransy theory based model offers a new method to better optimize the thermal management and gives specific measures to improve the thermal performance of data centers.

Evaluation metrics of thermal management in data centers based on exergy analysis

Cooling system has caused huge energy consumption in high-density data centers (DCs). Efficient thermal management makes a significant impact on energy saving of air conditioning system and safe operation of IT equipment. Air mixing in air flow space is an irreversible process leading to exergy loss from the view of Second-Law of Thermodynamics. In this paper, new exergetic evaluation metrics are proposed at room level and rack level respectively to reflect the causes and locations of inefficiency of cooling system in DCs. The effectiveness of new metrics is identified in a typical DC. Towards this, experimental measurement and numerical simulation are carried out to obtain the air distribution of thermal environment in a practical DC. Subsequently, the comparison with existing metrics demonstrates the applicability and advantage of new exergetic metrics. Exergy Loss Index below 50% is considered as good performance and it is suggested to be controlled below 80%. Finally, the contributions as well as the limitations of new proposed metrics are discussed when used in design and operation stages. It shows that introducing new exergetic metrics to the existing evaluation system is a more comprehensive way to evaluate the thermal management in typical DCs.

Experimental investigation on thermal management for small container data center

The energy consumptions of data centers are increasing dramatically and cooling systems using for controlling the operating temperature usually consume more than half of these energy. Thus, effective strategies on thermal management in data centers may have benefit on energy savings. In order to improve the power usage effectiveness (PUE) of data centers, the effects of air supply flowrate by computer room air handler (CRAH), intake flowrate of rack cooling fans, air supply grilles and heat load distribution on the cooling performance of a typical small container data center having overhead air supply system are investigated in this paper. The rack cooling index (RCI) and supply heat index (SHI) are incorporated to evaluate the local and average cooling performance of data racks. It is found that the intake flowrate of rack cooling fans plays dominant role, which is suggested to be greater than the air supply flowrate in normal operation. Meanwhile, the volumetric flowrate distribution of grilles is studied. The additional front grille before entering the cold aisle may improve the RCI and SHI of front rack (rack A), in which severe hot-air recirculation may occur in typical operation. Conversely, additional rear grille at the exit of cold aisle has negative effect on cooling performance of the last rack (rack E). Based on the measurement of local temperature distribution, the non-uniform heat load distribution in racks is proposed. Result shows that the RCI can be improved to 64% if more power is placed at lower half part, which is eligible for energy saving. However, the non-uniform heat load distribution shows limited improvement on the rear rack (rack E).

Optimal Noncollapsing Space-Filling Designs for Irregular Experimental Regions

ABSTRACTSpace-filling and noncollapsing are two important properties in designing computer experiments. We study how the noncollapsing, space-filling designs for irregular experimental regions can be generated efficiently by the proposed metaheuristic methods. We solve this optimal design problem using variants of the discrete particle swarm optimization (DPSO) approaches. Numerical results, including an application in data center thermal management, are used to illustrate the performances of the proposed algorithms. Based on these numerical results, we assert that the most efficient approach is to reformulate the target optimal design problem as a constrained optimization problem and then use a modified DPSO to solve the constrained optimization problem.

A review of thermal management and innovative cooling strategies for data center

Within a data center, 52% of the electricity is used by the IT equipment, 38% by the cooling system, and 10% for the remaining equipment (electrical power distribution, UPS…). The cooling system is mandatory to maintain optimal air temperature and relative humidity and provide a safe environment for all the electronic equipment. This paper presents a review on thermal management in data centers and various potential cooling technologies developed respecting energy saving constraint. Numerous researchers have proposed different alternative cooling strategy and solutions to free air cooling. This article describes a state of the art cooling strategies of data centers. The open aisle configuration is the most often used technique to remove heat. It does not affect energy efficiency but increases the electricity consumption dedicated to air conditioning. The most promising cooling systems, capable of absorbing energy flux while decreasing electricity consumption and rising the energy efficiency, are identified and discussed. It includes free cooling, liquid cooling, two-phase technologies and building envelope.

Comprehensive study on the effects of plenum depths on air flow and thermal managements in data centers

A comprehensive numerical study of air flow and thermal management of data centers is conducted for different plenum depths, tiles perforation ratios and rack power densities. Temperature/air flow profiles and data centers thermal management are quantified using the performance measuring parameters SHI (Supply Heat Index), RHI (Return Heat Index) and RTI (Return Temperature Index). A CFD (Computational Fluid Dynamics) approach for air flow and temperature distribution simulation in the data centers is used. The results showed that increasing the plenum depth: (i) increases the uniformity of air flow rate in the data center, (ii) increases the possibility of recirculation of hot air at the racks tops and bypass of cold air at the sides of the terminal racks of the racks row, and (iii) changes the values of the measurable performance parameters to be outside the recommended ranges. The results recommend that a perforated tiles opening ratio of 25–30% and a plenum depth of 60 cm can be considered as the more suitable values for the data centers.

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

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

Optimization of data center cooling efficiency using reduced order flow modeling within a flow network modeling approach

Much of the energy use by data centers is attributed to cooling, so improving the thermal management of data centers can translate to direct and significant economic benefits. However, data centers are complex systems that contain a significant number of components or sub-systems (e.g., servers, fans, pumps, and heat exchangers) that must be considered in any synergistic data center thermal efficiency improvement effort. In this paper, we demonstrate a novel hybrid modeling strategy for efficient and accurate second-law efficiency optimization of a data center cooling system using a proper orthogonal decomposition (POD) airflow modeling scheme with flow network modeling for external components. This strategy is applied to a data center cell, demonstrating energy and exergy savings of up to 23% and 43%, respectively, for a data center with large server utilization.

Comprehensive Experimental and Computational Analysis of a Fully Contained Hybrid Server Cabinet

The rapid growth in the number of data centers combined with the high-density heat dissipation of computer and telecommunications equipment has made energy efficient thermal management of data centers a key research area. 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 while the traditional air cooling approach requires overprovisioning. 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. Here, a hybrid-cooled enclosed cabinet and all its internal components were characterized experimentally in steady-state mode (e.g., experimentally determined heat-exchanger effectiveness and IT characterization). Also, a comprehensive numerical model of the cabinet was developed and validated using the experimental data. The computational model employs full numerical modeling of the cabinet geometry and compact models to represent the servers and the air/water heat exchanger. The compact models were developed based on experimental flow and thermal characterization of the internal components. The cabinet level model has been used to simulate a number of operating scenarios relevant to data center applications such as the effect of air leakage within the cabinet. The effect of the air side and the water side failure of the cooling system on the IT performance were investigated experimentally. A comparison was made of the amount of time required to exceed the operating temperature limit for the two scenarios.

Effect of CRAC units layout on thermal management of data center

Comprehensive numerical studies of thermal management of data centers were presented by several investigators for different geometric and operating conditions of data centers. In the present work, a technical note regarding the effect of the computer room air conditioning (CRAC) units layout arrangements is presented. Two arrangements of CRAC units layouts are investigated; namely locating CRACs units in line with the racks row and locating the CRACs units perpendicular to the rack row. Temperature distributions, air flow characteristics particularly air recirculation and bypass and thermal management in data centers are evaluated in terms of the measureable overall performance parameters: supply/return heat indices (SHI/RHI) and return temperature indices (RTI). The results showed that locating CRAC units perpendicular to the racks row has the following effects: (i) enhances the uniformity of the air flow from the perforated tiles along the rack row, (ii) reduces the hot air recirculation at the ends racks of the row and the cold air bypass at the middle rack of the row and (iii) enhances the data center performance parameters RTI, SHI and RHI.

Online Updating of Computer Model Output Using Real-Time Sensor Data

Data center thermal management has become increasingly important because of massive computational demand in information technology. To advance the understanding of the thermal environment in a data center, complex computer models are extensively used to simulate temperature distribution maps. However, due to management policies and time constraints, it is not practical to execute such models in a real time fashion. In this article, we propose a novel statistical modeling method to perform real-time simulation by dynamically fusing a base, steady-state solution of a computer model, and real-time thermal sensor data. The proposed method uses a Kalman filter and stochastic gradient descent method as computational tools to achieve real-time updating of the base temperature map. We evaluate the performance of the proposed method through a simulation study and demonstrate its merits in a data center thermal management application. Supplementary materials for this article are available online.

Experimental study of solving thermal heterogeneity problem of data center servers

Desirable thermal management of data center requires uniform temperature distribution along the servers. Hot air recirculation and cold air bypass in data center leads to non-homogeneous cold air distribution along the servers of the racks which may lead to heterogeneous temperatures distribution along the servers. The present work aims to experimentally study the possibility of controlling these heterogeneous temperature distributions by controlling the cold air flow rates along the servers. A physical scaled data center model was used to conduct this investigation. The effectiveness of thermal management of the servers racks of the data centers has been expressed in terms of intake, rare and surface temperature distributions along the rack servers and the supply and return heat indices (commonly symbolized as SHI and RHI; respectively). Excessive tests were firstly performed under uniform servers fans speed (uniform air flow rates through the different servers). Then the air flow rates distributions along the racks servers has been changed by regulating the server’s fans speeds using different schemes of fans speeds regulations at different data centers power densities. It is concluded that a uniform increase of server’s flow rate from the bottom to the top of servers rack cabinet provides (i) the lowest temperature at both cooling aisle (around 10%) and exhaust aisles (around 5%), (ii) the best uniform surface temperature of all rack servers (as the standard deviation is reduced from 10 to around 2), and (iii) the best values of thermal management metrics (SHI and RHI) typically SHI is reduced by around 20% while RHI is increased by around 3% to approach the targeted values; 0.1 and 0.9, respectively.

Study of hot air recirculation and thermal management in data centers by using temperature rise distribution

One of the key challenges in reducing the energy consumption of data centers is to improve the efficiency of air conditioning units. The efficiency is increased when the temperature of the supply air rises; however, the hot air recirculation and hot spots may limit the rise of temperature of the supply air to ensure normal operation of servers. To solve this problem, the energy equation is modeled to separate the thermal influence of each server or rack, and the temperature rise distribution (TRD) is proposed to illustrate the recirculation efficiency in data centers. Based on the TRD, two recirculation ratios are suggested to evaluate the recirculation of each rack in the data center. An algorithm is suggested to optimize the power distribution among racks to minimize the maximum temperature inside the data center, reduce the number of hot spots, and provide more space for increasing the temperature of the supply air. The metrics and algorithm are tested and proved efficient by a computational fluid dynamics (CFD) case.

Feedback Control Scheduling in Energy-Efficient and Thermal-Aware Data Centers

This paper presents a model-predictive control-based scheduling strategy called ThermoRing to reduce cooling costs in data centers. ThermoRing makes use of an online feedback control mechanism to improve thermal management of energy-efficient clusters in a data center. ThermoRing aims at keeping the maximum inlet temperatures of the nodes under a redline temperature limit with little stability errors. Importantly, the ThermoRing approach is capable of dealing with emergency conditions (e.g., node fan shutdown and unexpected rising task arrival rates) by dynamically balancing load among the nodes. ThermoRing incorporates a heat distribution matrix to model the thermal characteristics of a data center housing cluster. ThermoRing is conducive to thermal management in data centers with high-scheduling performance and stability. Using a real-world online bookstore trace, we conduct extensive experiments to compare the performance of ThermoRing with three existing solutions (i.e., C-Oracle, Ad-hoc, and MinHR). The experimental results show that ThermoRing improved the system throughput by more than 10% under regular load conditions and by 40% in emergency cases. ThermoRing also significantly improves the energy efficiency of MinHR, which is a thermal-aware scheduler.

Numerical investigation and parametric study for thermal and energy management enhancements in data centers' buildings

A CFD numerical investigation of air flow and temperature distribution in data centers is presented for different operating and geometric conditions. Numerical experiments were conducted to parametrically study the effects of perforated tiles opening ratio, data center power density and locations of Computer Room Air Conditioning (CRAC) on the data center performance. Temperature distribution especially hot spot points, air flow characteristics particularly air recirculation and bypass and thermal management in data centers are evaluated in terms of the measureable overall performance parameters: supply/return heat indices (SHI/RHI), return temperature indices (RTI) and return cooling indices (RCI). Techniques of using cold aisle containments for enhancing the cooling effectiveness of data centers are investigated and evaluated. The results showed that (i) increasing the perforation ratio of perforated tails decreases the recirculation and bypass of air flow around the rack, (ii) increasing data centers' power density and supplied air flow rates increases the air bypass and air re-circulation, (iii) changing the spacing between the CRAC units and cold aisle slightly changes of air recirculation and bypass around the racks, and (iv) using appropriate cold aisle containment improves the performance of cooling of data centers, especially at high power density.

Experimental investigations of air conditioning solutions in high power density data centers using a scaled physical model

The widespread use of data centers, the dramatical increase of the data center power density and the need for improving cooling system efficiency to maintain reliable operation temperature and save cooling energy make the study of data center thermal management an urgent issue. In the current paper, three different configurations for thermal management solution of high power density data centers are investigated, compared and evaluated. A scaled physical model data center has been designed and constructed for the sake of the study using the theory of scale modeling of air flow experiments. The results showed that (i) by using aisle partition and aisle containment the rack inlet temperature can be reduced by 3–13% and 13–15.5% for aisle partition and aisle containment configurations, respectively; (ii) the intake air temperature reduction increases with increasing power density; and (iii) using aisle partitions and aisle containment with raised floor improves the data center cooling performance.

A Brief Overview of Recent Developments in Thermal Management in Data Centers

Data centers are mission critical facilities that typically contain thousands of data processing equipment, such as servers, switches, and routers. In recent years, there has been a boom in data center usage, leading their energy consumption to grow by about 10% a year continuously. The heat generated in these data centers must be removed so as to prevent high temperatures from degrading their reliability, which would cost additional energy. Therefore, precise and reliable thermal management of the data center environment is critical. This paper focuses on recent advancements in data center modeling and energy optimization. A number of currently available and developmental thermal management technology in data centers are broadly reviewed. Computational fluid dynamics (CFD) for raised-floor data centers, experimental measurements, containment systems, economizer cooling, hybrid cooling, and device level cooling are all thoroughly reviewed. The paper concludes with a summary and presents areas of potential future research, which are based on the holistic integration of workload prediction and allocation, and thermal management using smart control systems.