Thermal Management Of Data Centers Research Articles

Measurement of Air Flow Rate Sensitivity to the Differential Pressure Across a Server Rack in a Data Center

Presently, air cooling is the most common method of thermal management in data centers. In a data center, multiple servers are housed in a rack, and the racks are arranged in rows to allow cold air entry from the front (cold aisle) and hot air exit from the back (hot aisle), in what is referred as hot-aisle-cold-aisle (HACA) arrangement. If the racks are kept in an open room space, the differential pressure between the front and back of the rack is zero. However, this may not be true for some scenarios, such as, in the case of cold aisle containment, where the cold aisle is physically separated from the hot data center room space to minimize cold and hot air mixing. For an under-provisioned case (total supplied tile air flow rate < total rack air flow rate) the pressure in the cold aisle (front of the rack) will be lower than the data center room space (back of the rack). For this case, the rack air flow rate will be lower than the case without the containment. In this paper, we will present a methodology to measure the rack air flow rate sensitivity to differential pressure across the rack. Here, we use perforated covers at the back of the racks, which results in higher back pressure (and lower rack air flow rate) and the corresponding sensitivity of rack air flow rate to the differential pressure is obtained. The influence of variation and nonuniformity in the server fan speed is investigated, and it is observed that with consideration of fan laws, one can obtain results for different average fan speeds with reasonable accuracy. The measured sensitivity can be used to determine the rack air flow rate with variation in the cold aisle pressure, which can then be used as a boundary condition in computational fluid dynamics (CFD)/rapid models for data center air flow modeling. The measured sensitivity can also be used to determine the change in rack air flow rate with the use of different types of front/back perforated doors at the rack. Here, the rack air flow rate is measured using an array of thermal anemometers, pressure is measured using a micromanometer, and the fan speed is measured using an optical tachometer.

Analysis of transient and hysteresis behavior of cross-flow heat exchangers under variable fluid mass flow rate for data center cooling applications

Effective thermal management of data centers is an important aspect of reducing the energy required for the reliable operation of data processing and communications equipment. Liquid and hybrid (air/liquid) cooling approaches are becoming more widely used in today's large and complex data center facilities. Examples of these approaches include rear door heat exchangers, in-row and overhead coolers and direct liquid cooled servers. Heat exchangers are primary components of liquid and hybrid cooling systems, and the effectiveness of a heat exchanger strongly influences the thermal performance of a cooling system. Characterizing and modeling the dynamic behavior of heat exchangers is important for the design of cooling systems, especially for control strategies to improve energy efficiency. In this study, a dynamic thermal model is solved numerically in order to predict the transient response of an unmixed–unmixed crossflow heat exchanger, of the type that is widely used in data center cooling equipment. The transient response to step and ramp changes in the mass flow rate of both the hot and cold fluid is investigated. Five model parameters are varied over specific ranges to characterize the transient performance. The parameter range investigated is based on available heat exchanger data. The thermal response to the magnitude, time period and initial and final conditions of the transient input functions is studied in detail. Also, the hysteresis associated with the fluid mass flow rate variation is investigated. The modeling results and performance data are used to analyze specific dynamic performance of heat exchangers used in practical data center cooling applications.

Manufacturing Energy Analysis of a Microchannel Heat Exchanger for High-density Servers

The power consumed by servers transforms into heat dissipated within IT equipment racks, which has led to the challenge of thermal management in data centers. This challenge has increased with the emergence of high-density servers. As heat dissipated per unit volume increases, conventional cooling becomes less effective and less efficient in maintaining an adequate environment for IT equipment. Liquid cooling has been demonstrated as an effective strategy to provide a reliable environment for servers and to reduce the load on conventional cooling systems. While microchannel process technology (MPT)-based devices offer a space-efficient approach to liquid cooling of high-density servers, MPT device manufacturing has been shown to be cost and energy intensive. Manufacturing of an MPT heat exchanger to liquid cool the warm exiting air from server racks is discussed. Manufacturing process energy analysis is conducted and compared for photochemical machining and two joining methods: diffusion bonding and laser welding. A process model for laser welding is developed, which enhances energy analysis. It is found that laser welding results in significantly lower life cycle energy impact than diffusion bonding due to reduced process energy and improved yield.

A practical approach to spatio-temporal analysis

This paper introduces a spatio-temporal statistical analysis approach ap- propriate for monitoring or managing a physical system in which measurements are taken over dense time resolution but at sparse locations. The proposed approach is designed for implementation in an automated and ecient operation with manual intervention required only for scenario analysis. The method is based on a mod- eling framework for complex predictor-response and spatio-temporal relationships, and issues model-based prediction intervals. To accommodate varying practical situations, the method also includes an automated decision criterion for choosing between parametric and nonparametric spatial covariance models. The approach is illustrated using a data center thermal management problem.

Dynamic Analysis of Cross Flow Heat Exchangers in Data Centers Using Transient Effectiveness Method

Heat exchangers are important facilities used in data center cooling systems. The effectiveness of heat exchangers strongly influences the thermal performance of cooling systems. Rear door heat exchangers, in-row, overhead coolers, and fully contained cabinets are some examples of executing liquid and hybrid cooling systems used in data centers. A liquid to liquid heat exchanger is an important component of the coolant distribution unit, which supplies chilled water to the aforementioned heat exchangers. Computer room air handler units are also typically liquid to air cross flow heat exchangers. It is important to characterize the dynamic behavior of these heat exchangers for the thermal management of data centers, to improve the design of control strategies, and energy efficiency. The transient response of a 2-D unmixed-unmixed cross flow heat exchanger, with a finite wall capacitance, is investigated. The transient effectiveness concept is used to build the mathematical models that will be used to solve and analyze the transient problems involving cross flow heat exchangers. Several transient cases are investigated numerically for step, ramp, and exponential variations in the inlet temperature of the minimum capacity rate fluid, as well as step and ramp changes in the flow rate of hot and cold fluids. The transient effectiveness characteristics are tested in parametric studies. The results obtained show that transient effectiveness is a useful method of analyzing and predicting the dynamic performance of cross flow heat exchangers.

Nonuniform temperature boundary condition effects on data center cross flow heat exchanger dynamic performance

Abstract Cross flow heat exchangers are important components used in data center cooling systems. The thermal environment usually becomes very complex due to the cooling system design configuration, including individually operated IT equipment inside the data center. Nonuniform fluid inlet temperature conditions may become very common in the cross flow heat exchangers, and may impact the performance of these heat exchangers. In this study, cross flow heat exchanger performance is investigated accounting for the nonuniform input effect of fluid inlet temperature. Several types of nonuniform inlet temperature boundary condition model are developed, based on the several actual problems in data center thermal management. The detailed effect of these nonuniform temperature boundary conditions on both the steady state and transient responses of a crossflow heat exchanger are investigated. The combined effects of the physical model parameters with nonuniform temperature boundary conditions are also analyzed. Several specific cases related to the practical application situations of cross flow heat exchangers and thermal management in data centers are outlined and analyzed in this study.

Real-time monitoring and evaluation of energy efficiency and thermal management of data centers

The rapid growth of IT industry and the miniaturization of semiconductors have resulted in substantial increase in energy consumption, power density of IT equipment and, subsequently, heat dissipated from data center racks. Metrics have been proposed to overcome these energy efficiency and thermal management challenges. Measuring the performance of the data centers using a combination of wisely chosen metrics can increase the opportunity for considerable energy reduction. A variety of metrics developed and applied for such evaluation are reviewed herein. The energy and cooling efficiency of a small data center is then evaluated by applying several metrics. To perform the analysis, real-time monitoring of 25 parameters over a period of six weeks was performed through design and implementation of a wireless monitoring network. The results are analyzed and current energy efficiency and thermal management issues are discussed with respect to the relative effectiveness of the various metrics.

Discrete particle swarm optimization for constructing uniform design on irregular regions

Central composite discrepancy (CCD) has been proposed to measure the uniformity of a design over irregular experimental region. However, how CCD-based optimal uniform designs can be efficiently computed remains a challenge. Focusing on this issues, we proposed a particle swarm optimization-based algorithm to efficiently find optimal uniform designs with respect to the CCD criterion. Parallel computation techniques based on state-of-the-art graphic processing unit (GPU) are employed to accelerate the computations. Several two- to five-dimensional benchmark problems are used to illustrate the advantages of the proposed algorithms. By solving a real application in data center thermal management, we further demonstrate that the proposed algorithm can be extended to incorporate desirable space-filling properties, such as the non-collapsing property.

A dynamic compact thermal model for data center analysis and control using the zonal method and artificial neural networks

Full-scale data center thermal modeling and optimization using computational fluid dynamics (CFD) is generally an extremely time-consuming process. This paper presents the development of a velocity propagation method (VPM) based dynamic compact zonal model to efficiently describe the airflow and temperature patterns in a data center with a contained cold aisle. Results from the zonal model are compared to those from full CFD simulations of the same configuration. A primary objective of developing the compact model is real-time predictive capability for control and optimization of operating conditions for energy utilization. A scheme is proposed that integrates zonal model results for temperature and air flow rates with a proportional–integral–derivative (PID) controller to predict and control rack inlet temperature more precisely. The approach also uses an Artificial Neural Network (ANN) in combination with a Genetic Algorithm (GA) optimization procedure. The results show that the combined approach, built on the VPM based zonal model, can yield an effective real-time design and control tool for energy efficient thermal management in data centers.

Numerical and Experimental Study of the Effect of Underfloor Blockages on Data Center Performance

Due to the increase in computer rack equipment power in recent years, thermal management of data centers has become a challenging problem. Data center facilities with raised floor plenums are the most popular configuration from a thermal management perspective. Considerable ongoing research efforts focus on optimizing the room layouts and equipment design in order to achieve the desired cooling. However, the detrimental impact of underfloor blockages, which occur widely, is seldom addressed. These blockages often take the form of chiller pipes, cabling, and wires. They impede the flow of cold air from the air conditioning units and yield unpredictable and undesirable air flow patterns. In this paper the effect of such underfloor blockages on data center performance is characterized in detail. A representative data center is modeled using a commercial computational fluid dynamics code with typical underfloor blockages. Blockages are shown to have a significant impact on tile flow rates and rack inlet temperatures. Based on the detailed numerical study broad guidelines are presented on managing the underfloor blockages for improved data center performance. Established guidelines are experimentally validated on a different data center cell. A detailed comparison between the experimental and numerical results is presented. Based on the numerical and experimental study it is concluded that blockages if placed in “critical” path can potentially have a detrimental impact on data center performance. Case studies are presented where blockages in “safe” path will have a minimum effect on data center performance.

Adaptive Probability-Based Latin Hypercube Designs

Adaptive sampling is an effective method developed mainly for regular regions. However, experimental regions in irregular shapes are commonly observed in practice. Motivated by a data center thermal management study, a new class of adaptive designs is proposed to accommodate a specific type of irregular region. Because the adaptive procedure introduces biases into conventional estimators, several design-unbiased estimators are given for estimating the population mean. Efficient and easy-to-compute unbiased estimators are also introduced. The proposed method is applied to obtain an adaptive sensor placement plan to monitor and study the thermal distribution in a data center. All of the supplemental materials used in this work are available online.

A Statistical Approach to Thermal Management of Data Centers Under Steady State and System Perturbations

Temperature control for a large data center is both important and expensive. On the one hand, many of the components produce a great deal of heat, and on the other hand, many of the components require temperatures below a fairly low threshold for reliable operation. A statistical framework is proposed within which the behavior of a large cooling system can be modeled and forecast under both steady state and perturbations. This framework is based upon an extension of multivariate Gaussian autoregressive hidden Markov models (HMMs). The estimated parameters of the fitted model provide useful summaries of the overall behavior of and relationships within the cooling system. Predictions under system perturbations are useful for assessing potential changes and improvements to be made to the system. Many data centers have far more cooling capacity than necessary under sensible circumstances, thus resulting in energy inefficiencies. Using this model, predictions for system behavior after a particular component of the cooling system is shut down or reduced in cooling power can be generated. Steady-state predictions are also useful for facility monitors. System traces outside control boundaries flag a change in behavior to examine. The proposed model is fit to data from a group of air conditioners within an enterprise data center from the IT industry. The fitted model is examined, and a particular unit is found to be underutilized. Predictions generated for the system under the removal of that unit appear very reasonable. Steady-state system behavior also is predicted well.

Energy-Efficient Thermal-Aware Task Scheduling for Homogeneous High-Performance Computing Data Centers: A Cyber-Physical Approach

High-performance computing data centers have been rapidly growing, both in number and size. Thermal management of data centers can address dominant problems associated with cooling such as the recirculation of hot air from the equipment outlets to their inlets and the appearance of hot spots. In this paper, we show through formalization that minimizing the peak inlet temperature allows for the lowest cooling power needs. Using a low-complexity linear heat recirculation model, we define the problem of minimizing the peak inlet temperature within a data center through task assignment (MPIT-TA), consequently leading to minimal cooling-requirement. We also provide two methods to solve the formulation: Xlnt-GA, which uses a genetic algorithm, and Xlnt-SQP, which uses sequential quadratic programming. Results from small-scale data center simulations show that solving the formulation leads to an inlet temperature distribution that, compared to other approaches, is 2 degC to 5 degC lower and achieves about 20 to 30 percent cooling energy savings at common data center utilization rates. Moreover, our algorithms consistently outperform the minimize heat recirculation algorithm, a recirculation-reducing task placement algorithm in the literature.

Optimization of Global Data Center Thermal Management Workload for Minimal Environmental and Economic Burden

The rapid deployment of information and communications technology (ICT) across the globe has led to a network of high-density computer data centers to store, process and transmit information. These large-scale technology warehouses consume vast amounts of energy for running the compute infrastructure and auxiliary cooling resources. Recent literature has suggested the possibility of globally staggering compute workloads to take advantage of local climatic conditions as a means to reducing cooling energy costs. This paper further explores this premise by performing an in-depth analysis of the environmental and economic burden of managing the thermal infrastructure of a globally connected data center network. The paper examines a case study where the potential energy savings achievable by staggering workloads across arbitrarily chosen data centers in the U.S., India, and Russia are examined. The results show that the environmental benefit of such off-shoring is mostly dependent on the fuel mix of the grid to which the workload is transferred and the energy consumption in each location. Further, we show that dynamic optimization of the thermal workloads based on local weather patterns can reduce the environmental burden by up to 30%. The paper concludes with a detailed economic assessment. For the case study in this paper, we find that such global workload staggering can potentially reduce operational costs by nearly 35%.

Exergy Analysis of Data Center Thermal Management Systems

The modeling of recirculation patterns in air-cooled data centers is of interest to ensure adequate thermal management of computer racks at increased heat densities. Most metrics that describe recirculation are based exclusively on temperature inside the data center, and therefore fail to provide adequate information regarding the energy efficiency of the thermal infrastructure. This paper addresses this limitation through an exergy analysis of the data center thermal management system. The approach recognizes that the mixing of hot and cold streams in the data center airspace is an irreversible process and must therefore lead to a loss of exergy. Experimental validation in a test data center confirms that such an exergy-based characterization in the cold aisle reflects the same recirculation trends as suggested by traditional temperature-based metrics. Further, by extending the exergy-based model to include irreversibilities from other components of the thermal architecture, it becomes possible to quantify the amount of available energy supplied to the cooling system that is being utilized for thermal management purposes. The energy efficiency of the entire data center cooling system can then be collapsed into the single metric of net exergy loss. When evaluated against a ground state of the external ambience, this metric enables an estimate of how much of the energy emitted into the environment could potentially be harnessed in the form of useful work. Thus, this paper successfully demonstrates that the proposed exergy-based approach can provide a foundation upon which the data center cooling system can be simultaneously evaluated for thermal manageability and energy efficiency.

Strategies for effective use of exergy-based modeling of data center thermal management systems

As power densities in data centers quickly increase, the inefficiencies of yesterday are becoming costly data center thermal management problems today. One proposed method to address the inefficiencies of state-of-the-art data centers is to use the concept of exergy. To this end, earlier investigations have used a finite-volume, uniform-flow computer model to analyze exergy destruction as a means of identifying inefficiencies. For this type of exergy-based program to be a useful engineering tool, it should: (i) be easy to set up, viz. establish grid size and impose system parameters; (ii) have a formulation that is solvable and numerically stable; (iii) be executable in reasonable time on a workstation machine with typical processor speed and memory; and (iv) model the physics with acceptable accuracy. This investigation explored specific strategies for achieving these features. This work demonstrates that optimally chosen computational strategies do enhance the usefulness of an exergy-based analysis program as an engineering tool for evaluating the thermal performance of a data center.

Investigation of indoor climate and power usage in a data center

Thermal management of data centers is an important issue for many high technology companies. The power requirement to provide proper indoor climate in data centers is considerable. Thus, possibilities for energy savings and efficient electricity utilization are quite important. In this paper, the airflow and temperature patterns, as well as the electrical power requirement within a small data center, have been investigated. The power requirement within the data center is considerably high, due to an oversized air conditioning system. An infrared camera was used to visualize the airflow and temperature pattern, showing that cool air does not reach the upper levels of the racks, despite a very high air exchange rate. Point measurements of temperatures in a rack show that recirculation cells are present, causing accumulation of heat and improper cooling of electronic equipments. Thus, the chilled air is not distributed properly and consequently the cooling energy is not used effectively.

Review of Cooling Technologies for Computer Products

This paper provides a broad review of the cooling technologies for computer products from desktop computers to large servers. For many years cooling technology has played a key role in enabling and facilitating the packaging and performance improvements in each new generation of computers. The role of internal and external thermal resistance in module level cooling is discussed in terms of heat removal from chips and module and examples are cited. The use of air-cooled heat sinks and liquid-cooled cold plates to improve module cooling is addressed. Immersion cooling as a scheme to accommodate high heat flux at the chip level is also discussed. Cooling at the system level is discussed in terms of air, hybrid, liquid, and refrigeration-cooled systems. The growing problem of data center thermal management is also considered. The paper concludes with a discussion of future challenges related to computer cooling technology.

Efficient Thermal Management of Data Centers—Immediate and Long-Term Research Needs

Power dissipation within computer rooms or data centers has been steadily increasing over the past decade. With the spread of CMOS technology into microprocessors and memory in the 1980s and 1990s, water-cooled mainframe systems were largely supplanted by lower power air-cooled systems. These systems were typically stacked into 2-m-high racks for efficient use of expensive data center floor space. Data center environmental cooling infrastructures correspondingly evolved into designs that recirculate hot exhaust air from the computer systems into air-conditioning units. The air-conditioning units remove the heat and return the cool air back into the room in a closed-loop fashion. These air-cooled infrastructures are largely open, nonducted environments where hot and cold airstreams are free to mix. The evolution of microprocessor fabrication technology has enabled the construction of high-power processors. The push by business, academia, and consumers for greater processing speed has motivated the design of computer systems that enable the greatest number of processors, and, thus, the greatest processing power, per rack volume. This increase in microprocessor density places a great deal of strain on current computer room environmental control technology. Furthermore, the rate of increase in power density in the data center is outpacing that of HVAC technology improvements. Because of this, computer manufacturers are faced with the choice of either limiting system performance in favor of reduced power consumption or of providing customers with higher performance products that are impractical to deploy. In this paper, we will highlight some of the primary challenges with cooling high-power density data centers. We will demonstrate that existing environmental infrastructures have inherent inefficiencies that can be very costly, and we will explore alternatives. Additionally, the use of numerical modeling to diagnose problems with data center design and layout will be demonstrated, and limitations to its effective use will be discussed. Finally, the high-power densities involved have increased the need for a theoretical treatment of data center thermophysics. We will discuss this need in detail and will suggest ways in which it might be addressed. Throughout the paper, focus will be placed on future directions with the hope of instilling enthusiasm for further research and development by academia and industry in this particular area of HVAC&R that will soon reach a critical point in its continuing evolution.