Conference Paper

Energy Savings Through Hot and Cold Aisle Containment Configurations for Air Cooled Servers in Data Centers

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

The increased focus on green technologies and energy efficiency coupled with the insatiable desire of IT equipment customers for more performance has driven manufacturers to deploy energy efficient technologies in the data centers. This paper describes a technique to achieve significant energy savings by preventing the cold and hot air streams within the data center from mixing. More specifically, techniques will be described that will separate the cool supply air to the server racks and exhaust hot air that returns to the air conditioning units. This separation can be achieved by three types of containment systems — cold aisle containment, hot aisle containment, and server rack exhaust chimneys. The advantages and disadvantages of each technique will be outlined. To show the potential for energy efficiency improvements a case study in deploying a cold aisle containment solution for a 8944 ft2 data center will be presented. This study will show that 59% of the energy required for the computer room air conditioning (CRAC) units used in a traditional open type data center could be saved.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Accordingly, the efficiency of cooling is a major focus of efforts to reduce DC electricity consumption, with good practice regarding air management, cooling equipment operating conditions and selection of efficient equipment forming the basis of academic and governmental studies and best practice guidelines [21,22]. This paper focuses on air management, since recent studies have highlighted the potential for energy savings through improvements in this area [23][24][25][26][27][28][29][30][31][32][33]. As with air conditioning for thermal comfort, efforts to minimise DC air conditioning energy consumption must be balanced against the need to maintain the desired thermal conditions. ...
... The effect of aisle containment on DC cooling has been investigated experimentally by Arghode et al. [34] and in CFD simulations by numerous authors [26,31,32,35,47]. It is not possible to determine the total level of bypass in the contained data centres studied in Arghode et al.'s work since only the flow rates through the racks and ACUs were measured [34]. ...
... Other CFD simulations have been used to demonstrate the potential reductions in electricity consumption resulting from the implementation of aisle containment: Schmidt et al. [31] reported a reduction in E ACU of 59%, whilst Shrivastava et al. [32] found a 33% reduction in the electricity consumption of the cooling infrastructure. Schmidt et al. [31] and Shrivastava et al. [32] report bypass percentages within contained systems of 3.1 and 13% respectively, although neither disclose any information regarding the model detail governing bypass from the contained aisles, nor do they report the value of Dp CH [31,32]. ...
Article
Full-text available
A combination of laboratory experiments and a system model are used to carry out the first investigation into the potential for cold air to bypass IT equipment within data centres (DCs) employing aisle containment, and the effect of this bypass on DC electricity consumption. The laboratory experiments involved applying a differential pressure across commercially available server racks and aisle containment systems and measuring the resulting air flow. The potential to minimise bypass by sealing leakage paths and redesigning racks was investigated and quantified experimentally. A new system model is developed using a combination of manufacturer data, empirical relationships and experimental results to predict the impact of bypass on the power consumption of the various components of a DC’s cooling infrastructure. The results show that, at typical cold aisle pressures, as much as 20% of the supplied air may bypass servers by finding alternate paths through the server rack itself. This increases the required flow rate from air conditioning units (ACUs). The system model predicts that: (i) practical measures undertaken to reduce this bypass could reduce total power consumption by up to 8.8% and (ii) excessive pressure differentials across the containment system could also increase power consumption, by up to 16%.
... One increasingly common method for reducing bypass and recirculation is to install physical barriers separating the hot and cold air streams, known as aisle containment systems [9], [13], [14]. Experimental and numerical investigations have shown this approach to be effective in reducing electricity consumption and improving cooling effectiveness [8], [9], [15]- [18]. In DCs employing aisle containment, a positive pressure differential, ∆p CH , is usually maintained between the cold and hot aisles, in order to prevent recirculation [9] [17]. ...
... Numerical investigations have either neglected the potential for bypass within server racks (i.e. cold air passing through server racks whilst bypassing the server inlets) [8], [13], [15], [16], [19], or have incorporated this phenomenon without experimental calibration [17] [20]. One experimental study investigated the impact of aisle containment on cooling, but did not measure bypass within racks, effectively assuming that all air flow through racks passed through the servers [9]. ...
Conference Paper
Full-text available
Aisle containment is increasingly common in data centres, and is widely believed to improve efficiency and effectiveness of cooling. Investigations into the impacts of aisle containment on the behavior and power consumption of cooling infrastructure and servers have been limited. Nor has the impact of supply air conditions on these factors been extensively investigated. This work uses measurements of bypass in a test data centre and observations on server behavior in a wind tunnel, in conjunction with a system model, to investigate the efficiency with which computations can be undertaken in an aisle contained data centre, and how this is impacted by supply air conditions.
... environmental control, such that thermal regulatory failure 43 and related effects can be efficiently avoided. While existing 44 literature and industry best practices already encompass 45 computational simulation-aided design methodologies [23]-46 [26] and dynamic load-balancing techniques at various soft-ware and hardware abstraction levels [27], [28], the fusion 48 of different approaches have remained practically limited, 49 and the exploitation of smart networks also presents much 50 untapped potential [29], [30]. In this paper, we propose 51 a multi-pronged approach comprising a predictive infras- ...
Article
Full-text available
Data centers are mission-critical infrastructures. There are strict service level requirements and standards imposed on operators and maintainers to ensure reliable run-the-clock operation. In the context of thermal management and data hall environmental control, the formation of hot and cold spots around server cabinets are especially undesirable, and can result in overheating, lifespan reductions, and performance throttling in the former and condensation damage in the latter. In this paper, we present a comprehensive multi-pronged methodology in data center environmental control, comprising computational fluid dynamics (CFD) simulation-aided predictive design first-stage approach, and a complementing Internet of Things (IoT) reactive management system that autonomously monitors and regulates fluctuations in thermal parameters. The novel hybrid methodology is demonstrated on various test scenarios derived from real-world context, and prototypes of the IoT system have been experimentally validated. The approach is shown to be efficient in eliminating unfavourable environmental variations and provides an enhanced understanding of common design problems and respective mitigation measures.
... The authors concluded that using "low-cost" CAC reduces the cooling requirements by 20%. Schmidt et al. [128] used CFD simulations to estimate savings from the CRAC unit if a containment system is used. They conclude that 59% of the energy consumed in the cooling unit can be saved. ...
Article
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.
... The coefficients of thermal network model can be obtained via field measurements or Computational Fluid Dynamics (CFD) simulations. To avoid the inefficiency of open aisle, later studies use the cold aisle containment to mitigate the recirculation [37,38]. It is revealed that enclosed cold aisle has positive effect on uniform thermal distribution and minimizing cooling energy consumption [39]. ...
Article
Full-text available
Energy-hungry data centers attract researchers’ attention to energy consumption minimization–a challenge confronted by both industry and academia. To assure the server reliability, its instantaneous temperature is generally controlled within a preset threshold. Nonetheless, field studies indicated that the occasional violation of extreme temperature constraints hardly affect the system reliability in practice. Therefore, strictly restraining the server temperature may contribute to meaningless energy consumption. As a response to this limitation, this paper presents a dynamic control algorithm without violating the average temperature constraint. We formulate a “soft” Server Temperature-Constrained Energy Minimization (STCEM) problem, where the object function consists of IT and Heat Ventilation & Air Conditioner (HVAC) energy. Based on the Lyapunov optimization, two algorithms, i.e., linear and quadratic control policies, are proposed to approximately solve the STCEM problem. The non-negative weight parameter V is introduced to trade off the energy consumption against server temperature constraint. Furthermore, extensive simulations have been carried out to evaluate the system performance for the proposed controlled algorithms. The experimental results demonstrate that the quadratic control policy outperforms the linear counterpart on the STCEM problem. Specifically, the energy consumption and server temperature constraint are well balanced when V≈5000.
... Some previous computational investigations also reported the benefits of cold-aisle containment. For example, Schmidt et al. [8] and Gondipalli et al. [9] indicated that the deployment of cold-aisle containment can result in significant energy savings. Moreover, in the case of cooling failure, cold-aisle containment was suggested to result in reduced hot spot generation with time as compared to the case without containment. ...
Article
Several previous studies on liquid cooling of high-power density electronic servers have been conducted for different configurations of the liquid channels, operating conditions and coolant, pure liquid or using nanofluid. Most of these studies have been done numerically or analytically without comprehensive experimental validation and comparison. To fill this research gap, a comprehensive experimental study was conducted in the present work to evaluate and compare the thermal performance of electronic devices using liquid cooling system by passing water, Al2O3 or CuO nanofluid in different flow mini-channels configurations (Serpentine, parallel and wavy channels). The results of the present experimental investigations showed that (i) heat sinks with serpentine mini-channels has a better thermal performance than heat sinks with wavy and parallel mini-channels heat sinks, (ii) the wavy channel heat sinks has a better thermal performance than parallel channel heat sinks, (iii) the presence of the nanoparticles provides a decrease in the thermal resistance and an augmentation in the average Nusselt number compared with pure water for all the liquid channels configurations with better performance of CuO-water nanofluid compared with Al2O3-water nanofluid, and (iv) for serpentine channels at 120 W power density, the minimum and maximum increase in the Nusselt number is (4.49%–12.5%) for 0.1% vol. Al2O3-water nanofluid and (5.77%–16.01%) at 0.1% vol. CuO-water nanofluid. For the purpose of validations and assessments, the results of the present experimental work were compared with the results of the previous analytical/numerical work as well as with the cases of the previous experimental works.
Conference Paper
The data center industry has experienced significant growth over the last decade, mainly due to the increased use of the internet for our day to day activities such as e-commerce, social media, video streaming and healthcare. This growth in demand results in higher energy costs, as data centers can be energy intensive facilities. A significant portion of the energy used in data centers is for cooling purposes. Hence, it is one of the important areas of optimization to be addressed to create more efficient data centers. Amongst the many ways to increase data center efficiencies, airflow management is a key solution to many existing data centers. Fundamentally, there are three main schemes: hot-aisle containment, cold-aisle containment and exhaust chimney containment. This paper's focus is to experimentally characterize the following cold aisle configurations: open aisle, partially contained aisle and fully contained aisles. Experimental data presented to evaluate the effectiveness of the different configurations are rack inlet contour plots, tile and rack flow rates, pressure measurements, and server CPU temperatures.
Article
The data center industry has experienced significant growth over the last decade, mainly due to the increased use of the internet for our day to day activities such as e-commerce, social media, video streaming, and healthcare. This growth in demand results in higher energy costs, as data centers can be energy intensive facilities. A significant portion of the energy used in data centers is for cooling purposes. Hence, it is one of the important areas of optimization to be addressed to create more efficient data centers. Among the many ways to increase data center efficiencies, air flow management is a key solution to many existing data centers. Fundamentally, there are three main schemes: hot-aisle containment, cold-aisle containment, and exhaust chimney containment. This paper's focus is to experimentally characterize the following cold aisle configurations: open aisle, partially contained aisle, and fully contained aisles. Experimental data presented to evaluate the effectiveness of the different configurations are rack inlet contour plots, tile and rack flow rates, pressure measurements, and server central processing unit (CPU) temperatures.
Article
Cold aisle containment is used in air cooled data centers to minimize direct mixing between cold and hot air. Here, we present room level air flow field investigation for open, partially and fully contained cold aisles. Our previous investigation for rack level modeling has shown that consideration of momentum rise above the tile surface, due to acceleration of air through the pores, significantly improves the predictive capability as compared to the generally used porous jump model. The porous jump model only specifies a step pressure loss at the tile surface without any influence on the flow field. The momentum rise could be included by either directly resolving the tile's pore structure or by artificially specifying a momentum source above the tile surface. In the present work, a modified body force model is used to artificially specify the momentum rise above the tile surface. The modified body force model was validated against the experimental data as well as with the model resolving the tile pore geometry at the rack level and then implemented at the room level. With the modified body force model, much higher hot air entrainment and higher server inlet temperatures were predicted as compared to the porous jump model. Even when the rack air flow requirement is matched with the tile air flow supply, considerable hot air recirculation is predicted. With partial containment, where only a curtain at the top of the cold aisle is deployed and side doors are opened, improved cold air delivery is suggested.
Article
Cold aisle containment is used in raised floor, air cooled data centers to minimize direct mixing between the supplied cold air and the hot air exiting from the servers. The objective of such a system is to minimize the server inlet air temperatures. In this paper, large scale air temperature field measurements are performed to investigate the hot air entrainment characteristics in the cold aisle in both open and contained aisle conditions. Both under-provisioned and over-provisioned scenarios were examined. Thermal field measurements suggest significant improvement in the cold air delivery for the case with contained aisle as compared to open aisle. Even for an over-provisioned case with open aisle, hot air entrainment was observed from the aisle entrance; however, for the contained aisle condition, close to perfect cold air delivery to the racks was observed. For both under-provisioned and over-provisioned cases, the aisle containment tended to equalize the tile and rack air flow rates. Balance air is expected to be leaked into or out of the containment to makeup the flow rate difference for the contained aisle condition. The CFD modeling strategy at the aisle level is also discussed for open aisle condition. Our previous investigation for rack level modeling has shown that consideration of momentum rise above the tile surface improves the predictive capability as compared to the generally used porous jump model. The porous jump model only specifies a step pressure loss at the tile surface without any influence on flow field. The momentum rise above the tile surface was included using a modified body force model by artificially specifying a momentum source above the tile surface. The modified body force model suggested higher air entrainment and higher reach of cold air as compared to the porous jump model. The modified body force model was able to better capture hot air entrainment through aisle entrance and compared well with the experimental data for the end racks. The generally used porous jump model suggested lower hot air entrainment and under predicted the server inlet temperatures for end racks.
Chapter
One of the objectives of an effective air flow management scheme is to minimize hot and cold air mixing. Appropriate arrangement of racks, such as hot-aisle-cold-aisle (HACA) partially meets this requirement. However, due to the tendency of entrainment of surrounding room air by emerging cold air stream from perforated tiles, hot and cold air mixing is very difficult to eliminate in open aisle condition. Use of physical barriers, separating the hot and cold regions shows promise to minimize this problem. In this chapter we will discuss the thermal characteristics of contained cold aisles, and their influence on the energy savings of a data center.
Article
Air-cooled data centers still represent the majority of all data centers. This paper discusses the fundamental aspect of improving efficiency in these data centers, namely, airflow management. At the room and row level, containment systems offer many benefits for reducing computer room air handler fan power, as well as cooling system overhead. This paper explores the history of the development as well as current code and policy issues affecting the use or need of containment systems. Specific efficiency gains are cited, and extensive discussion of practical aspects of containment systems is given.
Article
Full-text available
This paper theoretically investigates the relationships among factors that affect the temperature rise of server racks and experimentally tests the influence of variable space contained arrangements on the thermal performance. To express the flow and heat transfer process of cold air in servers and analyze the critical factors affecting the temperature rise, a simplified mathematical model representing servers is developed using experimental results. A experiment is conducted within a modular data center which cold air is supplied from a raised floor. The experiment employed a variable space of cold aisle containment, and measured the resulting temperature rise, as well as pressure difference of racks and other parameters, in the simplified mathematical model. By comparing the experimental results and theoretical calculation, the theoretical model is proved to be reasonable and valid. The model predicts that the critical factors affecting the temperature rise of racks consist of static and dynamic pressure difference, total pressure of the fans, geometric structure, power consumption, resistance of doors, and opening area of servers. The result shows that the factor affected by the cold aisle contained system is the static pressure, while for the dynamic pressure difference, the contained architecture has a slight positive effect. Although the average temperature rise is quite decreased in the contained system, the static pressure distribution is non-uniform. A half-contained system which reduced contained space ratio to 50% is measured to cause a 22% increase of the static pressure difference, making a more uniform temperature distribution.
Article
The rapid global increase in energy consumption by data centres requires new improved cooling solutions and techniques to be developed and implemented. In a typical data centre, approximately a third of the total power consumption is needed for the cooling infrastructure, resulting in high power usage effectiveness (PUE) values. The main culprits of raised PUE are legacy air-cooled data centres, exhausting only low grade waste heat for which capture and re-use is challenging. This study investigates numerically the potential for energy recuperation by a server-level internal layout optimization for a hybrid air/liquid-cooled server. The approach combines multi-objective genetic algorithm (MOGA) and entropy generation minimization (EGM) techniques to incorporate the multiple objectives involved in solving this problem, and examines the cooling performance and waste heat recovery potential. In order to evaluate the potential for waste heat recovery, an extra entropy generation term Ṡ ΔT,ext is introduced, representing an air/liquid heat exchanger at the rear of the server. The effect of modifying the internal component layout on pressure drop and the outlet temperature profile are of primary interest, due to their direct impact on fan power usage and energy recuperation potential. The CFD model of the baseline configuration is validated using experimental pressure measurements conducted on a real blade server. The research demonstrates that a basic server layout optimization such as changing the memory module angles and spacing could enhance both the cooling effectiveness but also improve the potential for waste heat recovery from the air stream. The maximum reduction in entropy generation rate due to server layout optimization is 15%, while the outlet temperature uniformity can be improved by up to 42%.
Article
Airflow is crucial for air-cooled data centers. Its flow path and distribution influences the thermal environment and energy efficiency of raised-floor data centers. This paper provides a review of the topic including the aspects of airflow factors, numerical study, airflow performance metrics, and thermal optimization. Based on the multi-scale characteristics of the data center, the thermal environment is categorized into room-level, rack-level, and server-level environments. For the room-level thermal environment, the main factors include layout, raised floor plenum and ceiling height, and perforated tiles. For the rack level, the effects of the porosity ratio of rack door, airflow rate/temperature, server population, server arrangement and power density are considered. For the server level, airflow rate and server fan speed are investigated. Moreover, numerical studies have been widely employed to understand the thermal environment of data centers. The selections of simulation tool and the methods for simplifying and validating the models are key to predicting the data center's thermal behavior correctly. In addition, airflow performance metrics and multi-scale thermal optimization are summarized and discussed. This review aims to emphasize the importance of the airflow in data centers and thus serve a guiding reference for airflow design and energy efficiency in data centers. Some recommended topics for future research are also provided.
Article
Isolated cold aisle airflow distribution is a transitional form from non-isolated airflow distribution to closed cold aisle airflow distribution. With the increase of the power of racks, cooling failure may happen in the layout of isolated cold aisle. This paper presents the study on cooling performance of the racks which are improved through bottom ventilating reform and adjustment. Inlet/outlet air temperature and mass flow rate of the racks are investigated in detail under conditions of various bottom ventilated areas and various porosity of raised floor. The characteristics of airflow distribution are contrastively analyzed through calculating indexes of thermal environment of data centers. Results show that adequate ventilation through the bottom of the racks is good for improvement of the state of airflow distribution. There is an optimization range (0.1 m∼0.3 m and 0.05 m∼0.15 m respectively) of ventilated area at the bottom of the racks. And high porosity (above 50%) of the ventilated area can reduce the inlet and outlet temperature of the racks and the racks in different positions have a better temperature uniformity. In conclusion, bottom ventilation of racks is a feasible plan to improve airflow distribution, and schemes of ventilated area and porosity of corresponding raised floor should be designed respectively under consideration of layout of racks and AC.
Conference Paper
Enclosed aisle data centers with bypass recirculation have previously been shown to reduce significantly the cooling infrastructure's power consumption compared to a typical open-aisle configuration. Enclosed aisle configurations also facilitate IT load placement through so-called thermally aware, energy-optimized virtualization. The formidable problem of optimal workload placement, which has received increased interest by the IT industry and has been the subject of numerous investigations, becomes much simpler, almost trivial, when the aisles are enclosed. This is because there is no preference among identical servers in an enclosed aisle based on the thermal environment, which will be uniform in this case, making for a wide range of equivalent load placement possibilities within the data center, or at least those racks in the data center that share enclosed cold aisles. However, the reduction in IT load can have a significant impact on the energy consumption of the cooling infrastructure due to the off-design operation of equipment. The work presented in this paper uses an experimentally validated thermo-hydraulic model of the data center's cooling infrastructure, previously developed by the authors, to explore optimization possibilities in air-cooled, enclosed aisle data centers. The model is used to evaluate the total energy consumption of the data center's cooling infrastructure for data centers operated at reduced IT load. The analysis highlights the importance of reducing the total power required for moving the air within the CRACs, the plenum, and the servers, rather than focusing primarily or exclusively on reducing the refrigeration system's power consumption and shows the benefits of bypass recirculation in enclosed aisle configurations. Energy efficient operating strategies for data centers with enclosed aisles are given for a range of IT loads and ambient conditions.
ResearchGate has not been able to resolve any references for this publication.