Analysis of the UNH Data Center Using CFD Modeling
|
|
- Suzan Moody
- 8 years ago
- Views:
Transcription
1 Applied Math Modeling White Paper Analysis of the UNH Data Center Using CFD Modeling By Jamie Bemis, Dana Etherington, and Mike Osienski, Department of Mechanical Engineering, University of New Hampshire, Durham, NH January, 2009 Introduction In the Fall of 2008, Applied Math Modeling Inc., the UNH Data Center, and students from the UNH Mechanical Engineering Program collaborated on a project to improve the design and thermal efficiency of the UNH Data Center. The UNH Data Center is an off-campus facility that consists of approximately 2700 square feet of floor space, divided into three smaller rooms with a common pressurized subfloor plenum. The goal of the project was first to create a mathematical model of the data center thermal environment using the CoolSim CFD (computational fluid dynamics) application, and then to use the model to improve the cooling efficiency of the room, thereby reducing energy consumption and/or allowing more IT capacity using the existing cooling equipment. Temperature and air flow measurements were made in the data center to ensure that the model was accurately predicting the system behavior. Figure 1: 2D plan view of the UNH Data Center 2009 Applied Math Modeling Inc. 1
2 Figure 2: 3D model of the UNH Data Center UNH Data Center Layout The existing data center is composed of three main rooms, as shown in Figure 1. The four external walls are considered the domain boundary for the model. This analysis was concerned only with the elements that lie within this boundary, so components such as the HVAC system of the building was not a focus. Instead, attention was centered on how the system dynamics within the rooms can be optimized to achieve a better overall thermal performance. The UNH Data Center is separated into three small rooms labeled Operations Rooms 1, 2, and 3, which are used to house cooling and IT equipment. The design of this data center is a result of the conversion of space once used as a cold storage facility. The partitioning of the data center into separate rooms stems from its original use and poses some unique challenges for cooling. The main elements in a data center are the racks of servers, power supplies, computer room air conditioning (CRAC) units, and the mechanism for distributing the cold air to the heat-generating components in the room. Here, a raised floor mechanism is used, where cold air from the CRAC units is blown into the subfloor plenum, and perforated tiles are positioned around the equipment to allow the cold air to enter the room and cool the equipment. This data center also utilizes a passive return system, where the heated exhaust air from the racks is allowed to pass freely within the rooms before entering the CRAC returns on top of the cooling units. The perforated tiles are labeled by letter in Figure 1, and the racks by number. A few of the rack numbers are missing. This is due to 2009 Applied Math Modeling Inc. 2
3 the fact that the racks are currently empty, but may be populated in future scenarios. A base model for the data center was created and a simulation was run. The results were used to analyze the flow interactions between the rooms, the effectiveness of the raised floor and existing perforated tile arrangement, the flow paths through and around the servers, and the warm air return to the tops of the air conditioning units. were unknown, so initial estimates were based on worst case specifications supplied by the manufacturer (corresponding to the highest heat loads). The heat loads were then adjusted downward depending on the system configuration. For example, systems configured with half the maximum total memory and CPU cores were modeled at 50% of the manufacturer s total thermal load. The UNH data center has a couple of unique characteristics that are important aspects of the existing room configuration. First, the subfloor region is small, with a height of only 9 inches, and a quantity of cable trays and wires in the subfloor reduces the available volume in the plenum and obstructs the flow of supply air in places. Second, the data center was partitioned into smaller rooms for security reasons prior to its current use as a data center. To allow for better performance as a data center, perforated grills were placed in the upper sections of the walls separating the rooms to allow the hot exhaust air to travel freely to the closest CRAC unit. The data center is cooled by four Liebert CRAC units. Units 1, 3, and 4 are located in Operations Room 2, while Unit 2 is located in Operations Room 1. Table 1 outlines the specifications for each unit. The data center contains 39 server racks. The majority of the racks are located in Operations Room 2. The heat loads for the servers Table 1: CRAC Unit Specifications Creating the Model in CoolSim The methodology for modeling the data center was to first create a model in CoolSim based on the physical characteristics of the room. This process entailed taking inventory of all the equipment types and placements in the room including racks, power supplies, perforated tile locations, walls, under-floor obstructions, cable cutouts, and any other areas where air might be blocked or leak through the raised floor. A baseline CoolSim simulation was then performed to gain insight about the general dynamics of air movement, under floor pressure, and temperature distribution throughout the facility. Care was taken to not make this initial model too complicated in terms of geometry detail, but to include enough detail to obtain the basic flow characteristics. Once the baseline results were calculated, comparisons were made to actual measurements taken in the 2009 Applied Math Modeling Inc. 3
4 data center. Based on the comparison between the simulated results and measurements, additional details were added to the model to more closely represent the measured flow characteristics. Creating the data center in CoolSim was fairly easy and took only about 4 hours. The CoolSim model was built using details of the room geometry, cooling units, rack locations, IT equipment in the racks, and thermal load assumptions for the equipment. All the significant leaks (areas greater than 6 square inches) in the system were also represented in the model. The internal walls separating the rooms included properly positioned vents to allow the flow of return air back to the CRAC units. The vents were assumed to have 40 percent open area. The most time-consuming aspect of creating the model was acquiring the technical specifications for each piece of IT equipment, required to model the thermal load. Unfortunately, rack electrical load measurements were not feasible in the UNH facility, so load assumptions had to be made using IT manufacturing data. Once the locations and technical specifications for all of the equipment were collected, an average load per rack and corresponding rack flow rate were calculated and input into the software. Based on an assumed 20 F temperature rise across each rack, the flow rate was computed by Cool- Sim to be 156 CFM/kW of thermal load. A Data Center Assessment Report, available in the software to provide global totals of heat loads, cooling capacities, and flow rates was used to summarize the equipment in the room. The report is shown in Figure 3. According to this summary, the CRACs provide 49% more cooling capacity and 71% more air flow than is required by the current racks due to the recent retirement of older IT servers. Since one of the CRACs is almost twenty years old and another is ten years old, it is likely that they are not operating at their maximum rated values. The surplus of cooling capacity may therefore be an overestimate if some of the units are operating at a lower performance level. This will be discussed in a later section when the measured rack temperatures are compared to the simulation results. Note, however, that the assessment report is simply a summary and does not take the flow and temperature distribution into account. A CFD simulation is required to predict these distributions, so will provide a far more accurate representation of whether or not the CRACs can adequately cool the equipment, even if some are running at less than peak capacity. Modeling Assumptions A number of assumptions were made for the creation of the base model. First, because it was not possible to make actual load measurements on each equipment rack in the facility, the maximum rack thermal load was determined by researching the technical specifications for each server and summed for each rack. By comparing the actual configurations in the racks (based on server usage) with the maximum published configurations, an assumption was made that the actual racks were producing 60% of the maximum published heat loads. Second, an approximation was made in the base model regarding the room geometry. In 2009 Applied Math Modeling Inc. 4
5 CoolSim Data Center Assessment Report Copyright (c) 2008, Applied Math Modeling Inc. Specifications Room dimensions 46 X 64 X 8.5 feet Room floor area 2944 sq. ft. Supply plenum height 0.83 feet Number of downflow CRAC units 4 Number of rack rows 27 Total number of racks 36 Number of tile rows 12 Total number of tiles 27 Number of PDUs 4 Number of baffles 13 Number of cable cutout 6 Airflow Assessment Supply air flow rate from downflow CRACs CFM Demand air flow rate from racks CFM Total demand air flow rate CFM Supply air flow rate (% of demand air flow rate) 171% (71% excess of the demand air flow rate.) Estimated average flow rate through perforated tiles 660 CFM Estimated average pressure drop across the tiles lbf/ft2 Estimated rate of air leakage through cable cutouts 204 CFM (1% of the total supply plenum flow rate) Estimated net supply airflow rate through perf tiles CFM (70% excess of the demand flow rate from racks) Thermal Assessment Estimated heat density W/sq.ft. of room area Downflow CRAC cooling capacity kw Supply air total cooling capacity kw Total rack heat load kw Total PDU heat load 3.66 kw Total cooling load kw Supply air cooling capacity (% of total cooling load) 149% (49% excess than cooling demand) Estimated temperature rise of supply air F Estimated average temperature of hot air F (for 60 F supply air temperature) Figure 3: The CoolSim Data Center Assessment Report this model, it was assumed that the subfloor was entirely clear of obstructions such as wires and pipes. This approximation was subsequently dropped because it yielded poor agreement with measurements of flow distribution through the perforated floor tiles. This point will be discussed in a later section. Third, the cooling capacity for each CRAC unit was based on technical specifications obtained from the manufacturer. The Cool- Sim software contains a library of DataAire and Liebert CRACs, so that the user can select the make and model number from the list and have all the specifications automatically supplied. This option assumes that the units 2009 Applied Math Modeling Inc. 5
6 are operating at the manufacturer s rated performance levels, which may decline over time due to age. In addition to creating a model using the manufacturer s cooling capacities, a model was run with reduced cooling capacities, where the reductions were proportional to the age of the unit. Experimental Measurements Over 200 measurements were taken in the data center to provide insight into the system behavior. Measurements of airflow and temperature in the data center were done using a Mannix hand-held Air Master WDCFM8951 device. Flow rate measurements where determined by measuring the average air velocity approximately 6 inches above the perforated tile. Measurements were made at roughly 5 positions on each tile - in the center and at each of the corners - and the average velocity was then multiplied by the area of the floor tile to obtain the volumetric flow rate. This method is not nearly as accurate as when a hooded device such as a Velometer is used, but it does represent what most data centers would have available. Velocity measurements were also taken at three different heights for each of the racks and averaged for comparison to simulated data. The Air Master 8951 was also used for temperature measurements, which were made at three points on rack inlets and exits, and then averaged for comparison to the simulated values. Table 2: Comparison of simulated and measured total air flow rate in the room Due to time constraints and instrument limitations, the measurements were sparse. Ideally, a meshing scheme could have been constructed to gather many data points for each measurement, so that accurate averages could be found, and random data points could be discarded. However, this approach was not practical due to the time sensitivity of the project. Furthermore, the goal of taking measurements was to obtain a general idea of the system behavior, and ensure that the model was representative. Therefore, the exactness of the data was less of a priority than the general trending patterns. Simulation Results and Model Validation Once the model was complete, a baseline simulation was run, and the results were compared to measurements taken in the data center. A systematic analysis was performed to achieve correlation between the model and the actual system. First, the total air flow predicted by the simulation was compared to the total measured air flow. In Table 2, a comparison between the total measured and simulated flow rates shows very good agreement, with an error of just under 5%. The fact that the simulation does not overpredict the measured airflow suggests that the CRAC flow rates have not declined from their rated values with age. The next comparison was of the distribution of air flow through the perforated floor tiles. Figure 4 illustrates the Cool- Sim baseline results for flow rate along with the measured data. The comparison is 2009 Applied Math Modeling Inc. 6
7 Figure 4: Air flow distribution through perforated floor tiles through Tiles 12 and 13 (L and M) but underpredicts the flow through Tiles (O U). One possible cause for this discrepancy is the presence of obstructions in the underfloor region that had not been included in the base model. For example, there is a large 6 inch cable bundle running along the length of the room, and a smaller 3 inch bundle running along the width of the room. Obstructions of this size in the narrow plenum are likely candidates for altering the supply airflow patterns and impacting the flow distribution through the tiles. Table 3: Tile labels and corresponding numerical values used for Figures 4 and 5 In a revised CoolSim model, these obstructions were added to the supply plenum, as done for each tile in the room, identified by number, with numbers corresponding to the letters referenced in Figure 1 in Table 3. As can be seen, the simulated and measured values for most of the perforated tiles are not in very good agreement, with error that ranges from less than 1% to about 60%. The worst agreement appears to be for Tiles (L through U), with the exception of Tile 14 (N). The simulation overpredicts the flow Figure 5: Pipes added to the plenum as underfloor obstructions 2009 Applied Math Modeling Inc. 7
8 distribution would be in even better agreement with measurements. Another way of examining the role of underfloor obstructions is to compare the trajectories of the supply air with and without them. Pathline plots are used for this purpose. They represent the paths Figure 6: Air flow distribution through tiles with underfloor obstructions taken by imaginary fluid particles shown in Figure 5. The revised predictions and can be colored by any variable. In Figure 7, the pathline plots in the supply ple- for airflow distribution were in much better agreement overall with the measurements, as num, colored by velocity magnitude, are shown in Figure 6. The average absolute error dropped by 8 points. It was believed by compared for the case without (left) and with (right) underfloor obstructions. Clearly, even the team that had additional complexity been the addition of two relatively small objects added to the underflow region, the predicted can have a significant impact on the flow- Figure 7: Supply plenum pathlines without (left) and with (right) underfloor pipes 2009 Applied Math Modeling Inc. 8
9 field, and hence, the airflow distribution through the perforated tiles. Once the predicted airflow correlated adequately with measurements, comparisons of the simulated and measured rack inlet and exit temperatures were performed. In Figure 8 the inlet temperatures are shown to be in good agreement with data, with an overall average error of 13%. The most consistent error in the rack inlet temperatures was for the row located in Operations Room 3, housing Racks 1 through 7 and in Operations Room 2, for Racks 29 through 34. For all of these racks, the simula- Figure 8: Rack inlet temperatures Figure 9: Rack Outlet Temperatures 2009 Applied Math Modeling Inc. 9
10 tion underpredicts the inlet temperatures. The airflow through the tiles in Room 1 is in good agreement with measurements and is underpredicted for the tiles servicing Racks 29 through 34. This means that the likely reason for the underpredicted temperatures is the assumption that the CRACs are operating with maximum cooling capacity. The air being supplied by the CRACs in the model is too cold. Even if one CRAC is not capable of meeting its cooling target, its warmer supply air could mix with the other supply air in the plenum and impact the rack inlet temperatures. It is also possible that additional under floor obstructions contribute to this discrepancy, since they would contribute to thermal mixing and air distribution. A comparison of rack exit (outlet) temperatures was also performed, and the results are presented in Figure 9. The exit temperature error averaged 10%. For Racks 1 through 7 and 29 through 34, the exit temperatures are less than the measured values, as was true for the inlet temperatures on these racks. This enforces the suggestion that the error is due to the supply air being too cold. In some cases, CoolSim underpredicts the rack inlet temperature yet overpredicts the rack outlet temperature. Rack 23 is one example, and it can be concluded that the rack heat load in the model was assumed to be too high. For Rack 22, on the other hand, the inlet temperature prediction was too high and the exit temperature prediction too low, suggesting that the rack heat load was assumed to be too low. In CoolSim, the cooling capacity type of boundary condition for CRACs removes only as much heat as needed to maintain a userspecified thermostat temperature at the return. The cooling capacity that is set by the user is simply the maximum heat that the unit can remove from the return air. When the manufacturer s cooling capacities were Figure 10: Rack inlet temperatures computed using full and reduced cooling capacity ratings on CRAC Units 3 and Applied Math Modeling Inc. 10
11 used in the simulation, the actual heat removal was reported to be less than these values for all CRACs in the data center. Even so, an additional simulation was performed to examine the effect of aging on two of the units. CRAC Unit 3 (about 10 years old) was set to operate at 80% of its rated value and CRAC Unit 4 (about 20 years old) was set to operate at 70%. CRAC Units 1 and 2 were not changed. The impact of these reductions on the rack inlet temperatures can be seen in Figure 10, where the simulated values using the full and reduced CRAC cooling capacities are shown. These results show a small increase in the predicted inlet temperatures, with the overall error dropping to 11%. The increase is small because the CRACs do not need to operate at their peaks, given the cooling demands of the room. Even so, the units that operate at their full ratings need to work harder when the other units operate at reduced levels. In Table Table 4: Cooling capacity boundary conditions and resulting heat removal for CRACs modeled with their full ratings and with reduced ratings on Units 3 (80%) and 4 (70%) 4, the actual heat removed by each CRAC is compared to the cooling capacity boundary condition for the two scenarios. When Units 3 and 4 operate at reduced levels, Units 1 and 2 operate at increased levels. The division of labor changes because the diminished cooling capacity of Unit 3 is smaller than the heat it was called upon to remove when running at maximum value. This exercise identifies Unit 3 as a potential weakness in the data center. It is old, yet needs to operate at close to its maximum rated value. Even the newest CRACS, Units 1 and 2, could be operating at reduced levels. A simulation was performed in which all of the units were derated by 30%, and this reduced the error for Racks 1 through 7 by 30% compared to the fully rated case and the overall error for the inlet temperatures by 40% (from 13% to 8%). Recommendations for Improved Efficiency Once the validity of the model was established, the simulation results were studied to see if recommendations could be made to improve the data center s efficiency. A number of key problem areas were found. First, as shown above, the cooling load was not evenly distributed between the four air conditioning units, and Unit 3 may be overtaxed. By balancing the load among the CRACs, the data center would be better prepared for a potential failure of one of them. Another common problem that was observed is that some of the cold air coming out of the floor tiles travels directly to the 2009 Applied Math Modeling Inc. 11
12 Figure 11: Short-circuiting of the supply air to the CRAC returns (royal blue) CRAC returns without ever passing through the racks. This phenomenon is known as short-circuiting and is shown in Figure 11 below. Here, supply pathlines trace the trajectories of air from their entry into the room at the perforated floor tiles to their exit points, which should be at the rack inlets. The image shows that some of the trajectories go directly to the CRAC returns (shown in blue) in Operations Room 2. While shortcircuiting in general reduced efficiency, some is to be expected in the UNH Data Center, however, because there is an abundance of supply air. As IT equipment is added to the facility in the future, it will be important to make sure that short-circuiting does not keep the available supply air from reaching the racks. Finally, CRAC 2 in Operations Room 1 is located in an area with virtually no heat load other than a power supply. By either removing the surrounding walls or relocating the unit, it could be more effectively utilized. These main results, as well as some smaller details, are now providing the foundation for the alternative designs. Conclusions and Future Work During the past four months, the UNH data center has been studied and the behavior of the system analyzed. The design team conducted extensive research to develop an understanding of the causes behind the current data center operating characteristics. At this point, the system has been effectively modeled, so the dynamics of the system are well understood. The model can now be used to test design alternatives to see if the efficiency of the facility can be improved Applied Math Modeling Inc. 12
13 A few obstacles were encountered during the first phase of this project. First, the task of obtaining measurements from the existing data center was challenging due to time constraints and instrument limitations. For future measurements, additional velocity and temperature transducers should be purchased. Furthermore, a velometer would provide an improved method of capturing the true volume flow rate from the perforated floor tiles, compared to the current method, which is somewhat crude. During the creation of the CFD model, approximations were made when specifying the heat loads of the racks. The original heat loads were based on technical specifications supplied by the manufacturers; however these values represented worst case scenarios (with the highest possible heat loads) and were not necessarily representative of the real system. Using the room measurements as a guide, the team was able to adjust the heat loads of the racks until valid results were achieved. Through this process, the team developed a sense of engineering judgment that will be useful in later aspects of the project. Acknowledgements The UNH design team would like to thank a number of people who made this project possible. We would like to recognize the management team at the UNH Data center, for their enthusiasm and cooperation throughout this project, and for allowing us to spend significant time in the data center gathering data. We would like to thank Professor John McHugh for acting as an advisor to the team, and for his assistance with the mathematical model of the system. Additionally, we would like to give a special thanks to the support staff at Applied Math Modeling Inc., for the use of their modeling tools, and for their valued support throughout the semester. The next focus of the project will be on developing system improvements and modeling the increase in thermal load anticipated for the next couple of years. All of these results will be presented to the director of UNH Data Services for review Applied Math Modeling Inc. 13
Reducing the Annual Cost of a Telecommunications Data Center
Applied Math Modeling White Paper Reducing the Annual Cost of a Telecommunications Data Center By Paul Bemis and Liz Marshall, Applied Math Modeling Inc., Concord, NH March, 2011 Introduction The facilities
More informationUsing CFD for optimal thermal management and cooling design in data centers
www.siemens.com/datacenters Using CFD for optimal thermal management and cooling design in data centers Introduction As the power density of IT equipment within a rack increases and energy costs rise,
More informationVerizon SMARTS Data Center Design Phase 1 Conceptual Study Report Ms. Leah Zabarenko Verizon Business 2606A Carsins Run Road Aberdeen, MD 21001
Verizon SMARTS Data Center Design Phase 1 Conceptual Study Report Ms. Leah Zabarenko Verizon Business 2606A Carsins Run Road Aberdeen, MD 21001 Presented by: Liberty Engineering, LLP 1609 Connecticut Avenue
More informationFundamentals of CFD and Data Center Cooling Amir Radmehr, Ph.D. Innovative Research, Inc. radmehr@inres.com
Minneapolis Symposium September 30 th, 2015 Fundamentals of CFD and Data Center Cooling Amir Radmehr, Ph.D. Innovative Research, Inc. radmehr@inres.com Learning Objectives 1. Gain familiarity with Computational
More informationBenefits of. Air Flow Management. Data Center
Benefits of Passive Air Flow Management in the Data Center Learning Objectives At the end of this program, participants will be able to: Readily identify if opportunities i where networking equipment
More informationOptimizing Network Performance through PASSIVE AIR FLOW MANAGEMENT IN THE DATA CENTER
Optimizing Network Performance through PASSIVE AIR FLOW MANAGEMENT IN THE DATA CENTER Lylette Macdonald, RCDD Legrand Ortronics BICSI Baltimore 2011 Agenda: Discuss passive thermal management at the Rack
More informationAir, Fluid Flow, and Thermal Simulation of Data Centers with Autodesk Revit 2013 and Autodesk BIM 360
Autodesk Revit 2013 Autodesk BIM 360 Air, Fluid Flow, and Thermal Simulation of Data Centers with Autodesk Revit 2013 and Autodesk BIM 360 Data centers consume approximately 200 terawatt hours of energy
More informationThinking Outside the Box Server Design for Data Center Optimization
Thinking Outside the Box Server Design for Data Center Optimization Marie Ross, Senior Applications Engineer Tom Gregory, Consultant Engineer October 2013 The benefits of analyzing thermal performance
More informationHow To Improve Energy Efficiency Through Raising Inlet Temperatures
Data Center Operating Cost Savings Realized by Air Flow Management and Increased Rack Inlet Temperatures William Seeber Stephen Seeber Mid Atlantic Infrared Services, Inc. 5309 Mohican Road Bethesda, MD
More informationUnified Physical Infrastructure (UPI) Strategies for Thermal Management
Unified Physical Infrastructure (UPI) Strategies for Thermal Management The Importance of Air Sealing Grommets to Improving Smart www.panduit.com WP-04 August 2008 Introduction One of the core issues affecting
More informationManaging Cooling Capacity & Redundancy In Data Centers Today
Managing Cooling Capacity & Redundancy In Data Centers Today About AdaptivCOOL 15+ Years Thermal & Airflow Expertise Global Presence U.S., India, Japan, China Standards & Compliances: ISO 9001:2008 RoHS
More informationAisleLok Modular Containment vs. Legacy Containment: A Comparative CFD Study of IT Inlet Temperatures and Fan Energy Savings
WH I TE PAPE R AisleLok Modular Containment vs. : A Comparative CFD Study of IT Inlet Temperatures and Fan Energy Savings By Bruce Long, Upsite Technologies, Inc. Lars Strong, P.E., Upsite Technologies,
More informationW H I T E P A P E R. Computational Fluid Dynamics Modeling for Operational Data Centers
W H I T E P A P E R Computational Fluid Dynamics Modeling for Operational Data Centers 2 Executive Summary Improving Effectiveness of CFD Technology in Cooling of Data Centers IT managers continue to be
More informationHow To Run A Data Center Efficiently
A White Paper from the Experts in Business-Critical Continuity TM Data Center Cooling Assessments What They Can Do for You Executive Summary Managing data centers and IT facilities is becoming increasingly
More informationPower and Cooling for Ultra-High Density Racks and Blade Servers
Power and Cooling for Ultra-High Density Racks and Blade Servers White Paper #46 Introduction The Problem Average rack in a typical data center is under 2 kw Dense deployment of blade servers (10-20 kw
More informationEducation Evolution: Scalable Server Rooms George Lantouris Client Relationship Manager (Education) May 2009
Education Evolution: Scalable Server Rooms George Lantouris Client Relationship Manager (Education) May 2009 Agenda Overview - Network Critical Physical Infrastructure Cooling issues in the Server Room
More informationBoosting the Energy efficiency of a Server Room: experimental + virtual approach.
Boosting the Energy efficiency of a Server Room: experimental + virtual approach. Nowadays, the increase in computing power allows the successful applications of the CFD simulation methods in civil environments.
More informationData Center Power Consumption
Data Center Power Consumption A new look at a growing problem Fact - Data center power density up 10x in the last 10 years 2.1 kw/rack (1992); 14 kw/rack (2007) Racks are not fully populated due to power/cooling
More informationRAISED-FLOOR DATA CENTER: PERFORATED TILE FLOW RATES FOR VARIOUS TILE LAYOUTS
Paper Presented at ITHERM 2004 Ninth Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems June 1-4, Las Vegas, NV RAISED-FLOOR DATA CENTER: PERFORATED TILE FLOW RATES
More informationIMPROVING DATA CENTER EFFICIENCY AND CAPACITY WITH AISLE CONTAINMENT
DATA CENTER RESOURCES WHITE PAPER IMPROVING DATA CENTER EFFICIENCY AND CAPACITY WITH AISLE CONTAINMENT BY: STEVE HAMBRUCH EXECUTIVE SUMMARY Data centers have experienced explosive growth in the last decade.
More informationEnergy Efficient Data Center Design. Can Ozcan Ozen Engineering Emre Türköz Ozen Engineering
Energy Efficient Data Center Design Can Ozcan Ozen Engineering Emre Türköz Ozen Engineering 1 Bio Can Ozcan received his Master of Science in Mechanical Engineering from Bogazici University of Turkey in
More informationHPPR Data Center Improvements Projects Towards Energy Efficiency (CFD Analysis, Monitoring, etc.)
HPPR Data Center Improvements Projects Towards Energy Efficiency (CFD Analysis, Monitoring, etc.) Felipe Visbal / March 27, 2013 HPPR Data Center Technologies Team Services Data Centers: Thermal Assessments
More informationIntroducing Computational Fluid Dynamics Virtual Facility 6SigmaDC
IT Infrastructure Services Ltd Holborn Gate, 330 High Holborn, London, WC1V 7QT Telephone: +44 (0)20 7849 6848 Fax: +44 (0)20 7203 6701 Email: info@itisltd.co.uk www.itisltd.com Introducing Computational
More informationStrategies for Deploying Blade Servers in Existing Data Centers
Strategies for Deploying Blade Servers in Existing Data Centers By Neil Rasmussen White Paper #125 Revision 1 Executive Summary When blade servers are densely packed, they can exceed the power and cooling
More informationVirtual Data Centre Design A blueprint for success
Virtual Data Centre Design A blueprint for success IT has become the back bone of every business. Advances in computing have resulted in economies of scale, allowing large companies to integrate business
More informationBenefits of Cold Aisle Containment During Cooling Failure
Benefits of Cold Aisle Containment During Cooling Failure Introduction Data centers are mission-critical facilities that require constant operation because they are at the core of the customer-business
More informationIPACK2005-73273 DISTRIBUTED LEAKAGE FLOW IN RAISED-FLOOR DATA CENTERS
Proceedings of IPACK5 ASME InterPACK '5 July 17-, San Francisco, California, USA IPACK5-7373 DISTRIBUTED LEAKAGE FLOW IN RAISED-FLOOR DATA CENTERS Amir Radmehr Innovative Research, Inc. Plymouth, MN, USA
More informationApplication of CFD modelling to the Design of Modern Data Centres
Application of CFD modelling to the Design of Modern Data Centres White Paper March 2012 By Sam Wicks BEng CFD Applications Engineer Sudlows March 14, 2012 Application of CFD modelling to the Design of
More informationData center upgrade proposal. (phase one)
Data center upgrade proposal (phase one) Executive Summary Great Lakes began a recent dialogue with a customer regarding current operations and the potential for performance improvement within the The
More informationProblem Statement In order to satisfy production and storage requirements, small and medium-scale industrial
Problem Statement In order to satisfy production and storage requirements, small and medium-scale industrial facilities commonly occupy spaces with ceilings ranging between twenty and thirty feet in height.
More informationBRUNS-PAK Presents MARK S. EVANKO, Principal
BRUNS-PAK Presents MARK S. EVANKO, Principal Data Centers of the Future and the Impact of High Density Computing on Facility Infrastructures - Trends, Air-Flow, Green/LEED, Cost, and Schedule Considerations
More informationA Comparative Study of Various High Density Data Center Cooling Technologies. A Thesis Presented. Kwok Wu. The Graduate School
A Comparative Study of Various High Density Data Center Cooling Technologies A Thesis Presented by Kwok Wu to The Graduate School in Partial Fulfillment of the Requirements for the Degree of Master of
More informationEffect of Rack Server Population on Temperatures in Data Centers
Effect of Rack Server Population on Temperatures in Data Centers Rajat Ghosh, Vikneshan Sundaralingam, Yogendra Joshi G.W. Woodruff School of Mechanical Engineering Georgia Institute of Technology, Atlanta,
More informationImproving Data Center Energy Efficiency Through Environmental Optimization
Improving Data Center Energy Efficiency Through Environmental Optimization How Fine-Tuning Humidity, Airflows, and Temperature Dramatically Cuts Cooling Costs William Seeber Stephen Seeber Mid Atlantic
More informationUsing CFD for Data Center Design and Analysis
Applied Math Modeling White Paper Using CFD for Data Center Design and Analysis By Liz Marshall and Paul Bemis, Applied Math Modeling Inc., Concord, NH January, 2011 Introduction Computational fluid dynamics
More informationAPC APPLICATION NOTE #92
#92 Best Practices for Designing Data Centers with the InfraStruXure InRow RC By John Niemann Abstract The InfraStruXure InRow RC is designed to provide cooling at the row and rack level of a data center
More informationStatement Of Work Professional Services
Statement Of Work Professional Services Data Center Cooling Analysis using CFD Data Center Cooling Analysis Using Computational Fluid Dynamics Service 1.0 Executive Summary Table of Contents 1.0 Executive
More informationData Center Cooling & Air Flow Management. Arnold Murphy, CDCEP, CDCAP March 3, 2015
Data Center Cooling & Air Flow Management Arnold Murphy, CDCEP, CDCAP March 3, 2015 Strategic Clean Technology Inc Focus on improving cooling and air flow management to achieve energy cost savings and
More informationUnified Physical Infrastructure SM (UPI) Strategies for Smart Data Centers
Unified Physical Infrastructure SM (UPI) Strategies for Smart Data Centers Deploying a Vertical Exhaust System www.panduit.com WP-09 September 2009 Introduction Business management applications and rich
More informationData Center Components Overview
Data Center Components Overview Power Power Outside Transformer Takes grid power and transforms it from 113KV to 480V Utility (grid) power Supply of high voltage power to the Data Center Electrical Room
More informationEnclosure and Airflow Management Solution
Enclosure and Airflow Management Solution CFD Analysis Report Des Plaines, Illinois December 22, 2013 Matt Koukl- DECP Mission Critical Systems Affiliated Engineers, Inc. Contents Executive Summary...
More informationThe New Data Center Cooling Paradigm The Tiered Approach
Product Footprint - Heat Density Trends The New Data Center Cooling Paradigm The Tiered Approach Lennart Ståhl Amdahl, Cisco, Compaq, Cray, Dell, EMC, HP, IBM, Intel, Lucent, Motorola, Nokia, Nortel, Sun,
More informationOptimum Climate Control For Datacenter - Case Study. T. Prabu March 17 th 2009
Optimum Climate Control For Datacenter - Case Study T. Prabu March 17 th 2009 Agenda 2 About EDEC (Emerson) Facility Data Center Details Design Considerations & Challenges Layout Design CFD Analysis Of
More informationDealing with Thermal Issues in Data Center Universal Aisle Containment
Dealing with Thermal Issues in Data Center Universal Aisle Containment Daniele Tordin BICSI RCDD Technical System Engineer - Panduit Europe Daniele.Tordin@Panduit.com AGENDA Business Drivers Challenges
More informationTen Steps to Solving Cooling Problems Caused by High- Density Server Deployment
Ten Steps to Solving Cooling Problems Caused by High- Density Server Deployment By Peter Hannaford White Paper #42 Revision 1 Executive Summary High-density servers present a significant cooling challenge.
More informationAPC APPLICATION NOTE #112
#112 Best Practices for Deploying the InfraStruXure InRow SC By David Roden Abstract The InfraStruXure InRow SC (ACSC100 and ACSC101) is a self-contained air conditioner for server rooms and wiring closets.
More informationElement D Services Heating, Ventilating, and Air Conditioning
PART 1 - GENERAL 1.01 OVERVIEW A. This section supplements Design Guideline Element D3041 on air handling distribution with specific criteria for projects involving design of a Data Center spaces B. Refer
More informationDataCenter 2020: hot aisle and cold aisle containment efficiencies reveal no significant differences
DataCenter 2020: hot aisle and cold aisle containment efficiencies reveal no significant differences November 2011 Powered by DataCenter 2020: hot aisle and cold aisle containment efficiencies reveal no
More informationCooling Audit for Identifying Potential Cooling Problems in Data Centers
Cooling Audit for Identifying Potential Cooling Problems in Data Centers By Kevin Dunlap White Paper #40 Revision 2 Executive Summary The compaction of information technology equipment and simultaneous
More information- White Paper - Data Centre Cooling. Best Practice
- White Paper - Data Centre Cooling Best Practice Release 2, April 2008 Contents INTRODUCTION... 3 1. AIR FLOW LEAKAGE... 3 2. PERFORATED TILES: NUMBER AND OPENING FACTOR... 4 3. PERFORATED TILES: WITH
More informationUnique Airflow Visualization Techniques for the Design and Validation of Above-Plenum Data Center CFD Models
Unique Airflow Visualization Techniques for the Design and Validation of Above-Plenum Data Center CFD Models The MIT Faculty has made this article openly available. Please share how this access benefits
More informationChoosing Close-Coupled IT Cooling Solutions
W H I T E P A P E R Choosing Close-Coupled IT Cooling Solutions Smart Strategies for Small to Mid-Size Data Centers Executive Summary As high-density IT equipment becomes the new normal, the amount of
More informationRittal Liquid Cooling Series
Rittal Liquid Cooling Series by Herb Villa White Paper 04 Copyright 2006 All rights reserved. Rittal GmbH & Co. KG Auf dem Stützelberg D-35745 Herborn Phone +49(0)2772 / 505-0 Fax +49(0)2772/505-2319 www.rittal.de
More informationHigh Density Data Centers Fraught with Peril. Richard A. Greco, Principal EYP Mission Critical Facilities, Inc.
High Density Data Centers Fraught with Peril Richard A. Greco, Principal EYP Mission Critical Facilities, Inc. Microprocessors Trends Reprinted with the permission of The Uptime Institute from a white
More informationGreat Lakes Data Room Case Study
Great Lakes Data Room Case Study WeRackYourWorld.com Problem: During a warm summer period in 2008, Great Lakes experienced a network outage due to a switch failure in the network enclosure. After an equipment
More informationFAC 2.1: Data Center Cooling Simulation. Jacob Harris, Application Engineer Yue Ma, Senior Product Manager
FAC 2.1: Data Center Cooling Simulation Jacob Harris, Application Engineer Yue Ma, Senior Product Manager FAC 2.1: Data Center Cooling Simulation Computational Fluid Dynamics (CFD) can be used to numerically
More informationLEVEL 3 DATA CENTER ASSESSMENT
Nationwide Services Corporate Headquarters 410 Forest Street Marlborough, MA 01752 USA Tel: 800-342-5332 Fax: 508-303-0579 www.eecnet.com LEVEL 3 DATA CENTER ASSESSMENT Submitted by: Electronic Environments
More informationHow Row-based Data Center Cooling Works
How Row-based Data Center Cooling Works White Paper 208 Revision 0 by Paul Lin and Victor Avelar Executive summary Row-based data center cooling is normally regarded as a cold air supply architecture that
More informationModeling Rack and Server Heat Capacity in a Physics Based Dynamic CFD Model of Data Centers. Sami Alkharabsheh, Bahgat Sammakia 10/28/2013
Modeling Rack and Server Heat Capacity in a Physics Based Dynamic CFD Model of Data Centers Sami Alkharabsheh, Bahgat Sammakia 1/28/213 2 ES2 Vision To create electronic systems that are self sensing and
More informationUsing Simulation to Improve Data Center Efficiency
A WHITE PAPER FROM FUTURE FACILITIES INCORPORATED Using Simulation to Improve Data Center Efficiency Cooling Path Management for maximizing cooling system efficiency without sacrificing equipment resilience
More informationUsing Simulation to Improve Data Center Efficiency
A WHITE PAPER FROM FUTURE FACILITIES INCORPORATED Using Simulation to Improve Data Center Efficiency Cooling Path Management for maximizing cooling system efficiency without sacrificing equipment resilience
More informationPrediction Is Better Than Cure CFD Simulation For Data Center Operation.
Prediction Is Better Than Cure CFD Simulation For Data Center Operation. This paper was written to support/reflect a seminar presented at ASHRAE Winter meeting 2014, January 21 st, by, Future Facilities.
More informationHow High Temperature Data Centers & Intel Technologies save Energy, Money, Water and Greenhouse Gas Emissions
Intel Intelligent Power Management Intel How High Temperature Data Centers & Intel Technologies save Energy, Money, Water and Greenhouse Gas Emissions Power and cooling savings through the use of Intel
More informationHow High Temperature Data Centers & Intel Technologies save Energy, Money, Water and Greenhouse Gas Emissions
Intel Intelligent Power Management Intel How High Temperature Data Centers & Intel Technologies save Energy, Money, Water and Greenhouse Gas Emissions Power savings through the use of Intel s intelligent
More informationStatement Of Work. Data Center Power and Cooling Assessment. Service. Professional Services. Table of Contents. 1.0 Executive Summary
Statement Of Work Professional Services Data Center Power and Cooling Assessment Data Center Power and Cooling Assessment Service 1.0 Executive Summary Table of Contents 1.0 Executive Summary 2.0 Features
More informationRack Hygiene. Data Center White Paper. Executive Summary
Data Center White Paper Rack Hygiene April 14, 2011 Ed Eacueo Data Center Manager Executive Summary This paper describes the concept of Rack Hygiene, which positions the rack as an airflow management device,
More informationReducing Room-Level Bypass Airflow Creates Opportunities to Improve Cooling Capacity and Operating Costs
WHITE PAPER Reducing Room-Level Bypass Airflow Creates Opportunities to Improve Cooling Capacity and Operating Costs By Lars Strong, P.E., Upsite Technologies, Inc. 505.798.000 upsite.com Reducing Room-Level
More informationUnderstanding How Cabinet Door Perforation Impacts Airflow by Travis North
& design deployment The first step in determining the proper perforation for an application is to understand the airflow requirements of the equipment that will be used in the cabinet. Understanding How
More informationGUIDE TO ICT SERVER ROOM ENERGY EFFICIENCY. Public Sector ICT Special Working Group
GUIDE TO ICT SERVER ROOM ENERGY EFFICIENCY Public Sector ICT Special Working Group SERVER ROOM ENERGY EFFICIENCY This guide is one of a suite of documents that aims to provide guidance on ICT energy efficiency.
More informationAirflow Simulation Solves Data Centre Cooling Problem
Airflow Simulation Solves Data Centre Cooling Problem The owner s initial design for a data centre in China utilized 40 equipment racks filled with blade servers spread out in three rows along the length
More informationControl Strategies for Plenum Optimization in Raised Floor Data Centers
Control Strategies for Plenum Optimization in aised Floor Data Centers Keke Chen, Clifford C. Federspiel, David M. Auslander, Cullen E. Bash, Chandrakant D. Patel Enterprise Software and Systems Laboratory
More informationSaving energy in the Brookhaven National Laboratory Computing Center: Cold aisle containment implementation and computational fluid dynamics modeling
Saving energy in the Brookhaven National Laboratory Computing Center: Cold aisle containment implementation and computational fluid dynamics modeling Student Intern Earth and Environmental Engineering,
More information11 Top Tips for Energy-Efficient Data Center Design and Operation
11 Top Tips for Energy-Efficient Data Center Design and Operation A High-Level How To Guide M e c h a n i c a l a n a l y s i s W h i t e P a p e r w w w. m e n t o r. c o m When is Data Center Thermal
More informationAEGIS DATA CENTER SERVICES POWER AND COOLING ANALYSIS SERVICE SUMMARY
AEGIS DATA CENTER SERVICES POWER AND COOLING ANALYSIS SERVICE SUMMARY The Aegis Services Power and Assessment Service provides an assessment and analysis of your data center facility and critical physical
More informationEnvironmental Data Center Management and Monitoring
2013 Raritan Inc. Table of Contents Introduction Page 3 Sensor Design Considerations Page 3 Temperature and Humidity Sensors Page 4 Airflow Sensor Page 6 Differential Air Pressure Sensor Page 6 Water Sensor
More informationMEASUREMENTS AND PREDICTIONS OF THE FLOW DISTRIBUTION THROUGH PERFORATED TILES IN RAISED-FLOOR DATA CENTERS
Proceedings of IPACK 01 The Pacific Rim/ASME International Electronic Packaging Technical Conference and Exhibition July 8 13, 2001, Kauai, Hawaii, USA IPACK2001-15728 MEASUREMENTS AND PREDICTIONS OF THE
More informationFlow in data racks. 1 Aim/Motivation. 3 Data rack modification. 2 Current state. EPJ Web of Conferences 67, 02070 (2014)
EPJ Web of Conferences 67, 02070 (2014) DOI: 10.1051/ epjconf/20146702070 C Owned by the authors, published by EDP Sciences, 2014 Flow in data racks Lukáš Manoch 1,a, Jan Matěcha 1,b, Jan Novotný 1,c,JiříNožička
More informationCooling Capacity Factor (CCF) Reveals Stranded Capacity and Data Center Cost Savings
WHITE PAPER Cooling Capacity Factor (CCF) Reveals Stranded Capacity and Data Center Cost Savings By Lars Strong, P.E., Upsite Technologies, Inc. Kenneth G. Brill, Upsite Technologies, Inc. 505.798.0200
More informationSupporting Cisco Switches In Hot Aisle/Cold Aisle Data Centers
CABINETS: ENCLOSED THERMAL MOUNTING MANAGEMENT SYSTEMS WHITE PAPER Supporting Cisco Switches In Hot Aisle/Cold Aisle Data Centers 800-834-4969 techsupport@chatsworth.com www.chatsworth.com All products
More informationComputational Fluid Dynamic Investigation of Liquid Rack Cooling in Data Centres
School of something School of Mechanical Engineering FACULTY OF OTHER ENGINEERING Computational Fluid Dynamic Investigation of Liquid Rack Cooling in Data Centres A. Almoli 1, A. Thompson 1, N. Kapur 1,
More informationAirflow Utilization Efficiency (AUE) Can we cool today s servers with airflow?
AirflowUtilizationEfficiency(AUE) How to use River Cooling to reduce energy costs while improving server performance. By ThomasS.Weiss Sr.VicePresident TriadFloorsbyNxGen,LLC This paper offers insight
More informationEnergy Management Services
February, 2012 Energy Management Services Data centers & IT environments are often critical to today s businesses, directly affecting operations and profitability. General & Mechanical Services has the
More informationRaised Floor Data Centers Prepared for the Future
Raised Floor Data Centers Prepared for the Future Raised Floors Protect My Investment by Providing a Foundation for Current and Future Technologies. Creating the Future Proof Data Center Environment The
More informationData Centers and Mission Critical Facilities Operations Procedures
Planning & Facilities Data Centers and Mission Critical Facilities Operations Procedures Attachment A (Referenced in UW Information Technology Data Centers and Mission Critical Facilities Operations Policy)
More informationPrompt Engineering applies successfully CFD analysis on large Server Room for a simulation driven Energy saving approach.
Prompt Engineering applies successfully CFD analysis on large Server Room for a simulation driven Energy saving approach.. ABSTRACT The Energy Saving approach is the target of this case story. The problem
More informationImproving Rack Cooling Performance Using Airflow Management Blanking Panels
Improving Rack Cooling Performance Using Airflow Management Blanking Panels By Neil Rasmussen White Paper #44 Revision 3 Executive Summary Unused vertical space in open frame racks and rack enclosures
More informationData Center Airflow Management Retrofit
FEDERAL ENERGY MANAGEMENT PROGRAM Data Center Airflow Management Retrofit Technology Case Study Bulletin: September 2010 Figure 1 Figure 2 Figure 1: Data center CFD model of return airflow short circuit
More informationRole of the PI System in Research @ Georgia Tech Data Center Laboratory
Role of the PI System in Research @ Georgia Tech Data Center Laboratory Presented by Vaibhav K. Arghode Prepared by Vikneshan Sundaralingam Outline Data center laboratory general information Current use
More informationIT Cost Savings Audit. Information Technology Report. Prepared for XYZ Inc. November 2009
IT Cost Savings Audit Information Technology Report Prepared for XYZ Inc. November 2009 Executive Summary This information technology report presents the technology audit resulting from the activities
More informationLiquid Cooling Solutions for DATA CENTERS - R.M.IYENGAR BLUESTAR LIMITED.
Liquid Cooling Solutions for DATA CENTERS - R.M.IYENGAR BLUESTAR LIMITED. Presentation Goals & Outline Power Density Where we have been- where we are now - where we are going Limitations of Air Cooling
More informationMobile Visual Analytics for Data Center Power and Cooling Management
Mobile Visual Analytics for Data Center Power and Cooling Management Ratnesh Sharma, Ming Hao, Ravigopal Vennelakanti, Manish Gupta, Umeshwar Dayal, Cullen Bash, Chandrakant Patel, Deepa Naik, A Jayakumar,
More informationSealing Gaps Under IT Racks: CFD Analysis Reveals Significant Savings Potential
TECHNICAL REPORT Sealing Gaps Under IT Racks: CFD Analysis Reveals Significant Savings Potential By Lars Strong, P.E., Upsite Technologies, Inc. Bruce Long, Upsite Technologies, Inc. +1.888.982.7800 upsite.com
More informationEnergy Efficiency Opportunities in Federal High Performance Computing Data Centers
Energy Efficiency Opportunities in Federal High Performance Computing Data Centers Prepared for the U.S. Department of Energy Federal Energy Management Program By Lawrence Berkeley National Laboratory
More informationIT White Paper MANAGING EXTREME HEAT: COOLING STRATEGIES FOR HIGH-DENSITY SYSTEMS
IT White Paper MANAGING EXTREME HEAT: COOLING STRATEGIES FOR HIGH-DENSITY SYSTEMS SUMMARY As computer manufacturers pack more and more processing power into smaller packages, the challenge of data center
More informationCase Study: Opportunities to Improve Energy Efficiency in Three Federal Data Centers
Case Study: Opportunities to Improve Energy Efficiency in Three Federal Data Centers Prepared for the U.S. Department of Energy s Federal Energy Management Program Prepared By Lawrence Berkeley National
More informationEnergy Performance Optimization of Server Room HVAC System
International Journal of Thermal Technologies E-ISSN 2277 4114 2 INPRESSCO, All Rights Reserved Available at http://inpressco.com/category/ijtt/ Research Article Manoj Jadhav * and Pramod Chaudhari Department
More informationCenter Thermal Management Can Live Together
Network Management age e and Data Center Thermal Management Can Live Together Ian Seaton Chatsworth Products, Inc. Benefits of Hot Air Isolation 1. Eliminates reliance on thermostat 2. Eliminates hot spots
More informationThermal Monitoring Best Practices Benefits gained through proper deployment and utilizing sensors that work with evolving monitoring systems
NER Data Corporation White Paper Thermal Monitoring Best Practices Benefits gained through proper deployment and utilizing sensors that work with evolving monitoring systems Introduction Data centers are
More informationDriving Data Center Efficiency Through the Adoption of Best Practices
Data Center Solutions 2008 Driving Data Center Efficiency Through the Adoption of Best Practices Data Center Solutions Driving Data Center Efficiency Through the Adoption of Best Practices Executive Summary
More informationData Centre Energy Efficiency Operating for Optimisation Robert M Pe / Sept. 20, 2012 National Energy Efficiency Conference Singapore
Data Centre Energy Efficiency Operating for Optimisation Robert M Pe / Sept. 20, 2012 National Energy Efficiency Conference Singapore Introduction Agenda Introduction Overview of Data Centres DC Operational
More information