Log In   |   Sign up

New User Registration

Article / Abstract Submission
Register here
Register
Press Release Submission
Register here
Register
coolingZONE Supplier
Register here
Register

Existing User


            Forgot your password
Sherman Ikemoto | April 2009

Using Simulation to Improve Data Center Efficiency


using simulation to improve data center efficiency

cooling path management for maximizing cooling system efficiency without sacrificing equipment resilience

akhil docca and sherman ikemoto

executive summary

modern it equipment with high power dissipation and powerful internal cooling fans are a primary cause of cooling and efficiency problems in today’s mission critical facility.these problems appear when high-velocity, high-temperature exhaust air from it equipment flows unintentionally into the inlets of surrounding equipment despite measures taken to prevent such occurrences.the conventional approaches to data center cooling design such as the hot aisle/cold aisle layout, kw per cabinet limits and the use of traditional computational fluid dynamics (cfd) software tools are blind to equipment related cooling and efficiency problems – a primary cause of data center efficiency problems.

this paper describes a new cfd-based modeling capability and design methodology that overcomes the limitations of conventional guidelines and traditional cfd-based approaches and is effective at addressing equipment related cooling and efficiency problems in the data center.depending on circumstances, 30-50% improvement in cooling system efficiency can be achieved by this specialized cfd-based approach.

the efficiency challenge of the modern data center

make more power available to the it equipment

much of the power used for data center cooling is wasted due to suboptimal airflow conditions such as supply air bypassing the it equipment and flowing directly to the air conditioning unit (acu) returns. this article will present a step-by-step process to analyze and optimize the paths that air takes from the acus to the equipment inlets and back. the process can be used to substantially improve cooling system efficiency by configuring any combination of it and cooling equipment to reduce bypass and recirculation.as a result, more power becomes available for use by the it equipment or to reduce energy bills. often efficiency improvements of 30-50% or more can be achieved by this simulation based design and optimization technique.

energy efficiency opportunities

power utilization in the data center is typically split 50 – 40 between the it equipment and the cooling system with about10% of total power used by various other equipment. [1]the actual percentage breakdown for any specific facility will vary but, in most cases the cooling system offers a significant opportunity to increase the efficiency that is under direct control of the owner/operator.cooling system power is typically split evenly between the fans and the chilled water system.cooling system efficiency, therefore, can be improved specifically by reducing fan and chilled water system power consumption.

<!--[if !mso]-->

<!--[endif]-->

figure 1: air recirculation within cabinets is a common cause of overheating and efficiency problems

<!--[if !mso]-->
<!--[endif]--> for historical reasons, cooling systems are often designed under sweeping assumptions that grossly oversimplify the physics of airflow.for example, it is common to use a maximum kw per cabinet to approximate the worst case it loading condition.there are three major problems with this type of sweeping assumption.

firstly, it is completely blind to the most common cooling problems that occur in the data center which are thermal interactions between the it equipment and the surrounding environment.an example is shown in figure 1 where exhaust air cannot escape the cabinet and is forced back into the intake through a gap between the bottom of the unit and the cabinet.

figure 1: air recirculation within cabinets is a common cause of overheating and efficiency problems

secondly, equipment related thermal problems cannot be solved for all-time with a single, blanket design solution because the it equipment is under constant change due to refresh or changing business requirements.over time, data centers experience dozens of different worst case conditions.each condition must be addressed individually at the time of occurrence to maintain acceptable levels of resilience and operating efficiency.

finally, a single, blanket cooling solution by definition leads to over cooling, excessive power consumption and reduced efficiency in order to cover all possible worst case conditions.often 30-50% of the power used to operate the chiller and the air handler fans can be recovered by a cooling design that fits the current it configuration. [2] fan power consumption is proportional to the cubic root of fan speed so reducing airflow by a factor of two reduces fan power consumption by a factor of eight.similarly, a 9 �f increase in chilled water temperature set point decreases chiller power consumption by about 25%.��

given these challenges, a design methodology is needed that overcomes the limitations of the conventional design guidelines (hot aisle/cold aisle, blanking) and the traditional cfd-tools by properly accounting for the cooling and efficiency issues caused by it equipment and the time varying nature of it deployments.

addressing the energy efficiency challenge

cooling path management

cooling path management (cpm) is a cfd-based methodology for data center cooling design and optimization that can address all possible cooling breakdowns and sources of inefficiency for any configuration of room and combination it equipment.cpm can be used for initial cooling system design and for ongoing management of cooling system performance that is a challenge due to unanticipated, localized problems caused by it equipment and the time-varying nature of it deployments.cpm is based on a specialized implementation of cfd analysis that solves the governing equations for fluid motion and heat transfer over the full range of physical scale that occurs in the data center (from small gaps within cabinets to the open room) and provides a 3d graphical view of the entire “cooling path” over which the breakdowns occur.

the cooling path is the route taken by the cooling air from the air handlers to the intakes of each unit of it equipment and back.cooling paths are formed by the thermal and physical characteristics of the equipment, cabinets and room and the way they are configured within the data center.the cooling path can be split into three primary segments to simplify the cpm design process as shown in figure 2.

each segment of the cooling path has a specific design goal and associated set of design options. this makes cpm intuitive, easy to use and applicable to all possible combinations of it equipment, cabinets and rooms.

key concept:a cooling path is 100% efficient when it is completely free of vortices, bypass and recirculation

the design objective for segment #1 is to meet the temperature and airflow requirement for each floor grille by finding the right combination of floor height, air handler selection and placement of air handlers, floor grilles and under floor objects.the design process is often streamlined by pre-setting some of the variables and then sequentially setting the remaining variables to meet the requirements.a common practice starts with a hot aisle/cold aisle floor tile arrangement followed by air handler selection and placement.fine tuning can be performed by adjusting the locations of the air handlers, under floor objects (pipes and cable trays) and floor grilles.cpm is an effective technique for under floor and airflow return design and is universally applicable to any configuration of room and floor grille arrangement.

the design objectives for segments #2 and #3 are to meet the intake temperature and airflow requirements for each unit of it equipment.additional requirements to minimize bypass and recirculating airflow can be imposed to fine tune to increase cooling system efficiency.

these requirements are met by finding the combination of cabinet design, cabinet placement, cable routing and u-slot location that are compatible with the thermal characteristics of the it equipment to be installed.cpm greatly streamlines design and optimization of segments #2 and #3 when large numbers of cabinets and it equipment are involved.

three properties of it equipment - power dissipation, cfm requirement and draft configuration combine to form a unique “thermal footprint” and become the source of cooling path formation for segments #2 and #3.figure 3 illustrates how thermal footprints are unique to it equipment model type.

thermal footprints interact with the surrounding environment to define cooling path segments #2 and #3 that lead to and away from the equipment inlets and exhausts as illustrated in figure 4.

figure 4:�thermal footprints interact with the surrounding environment to form segments #2 and #3 of cooling path

key concept:understanding the interactions between thermal footprint and the environment is the basis for addressing the most common data center cooling and efficiency problems

given the impact of thermal footprints on the data center environment, it is no surprise that cooling path segments #2 and #3 can change dramatically with each configuration of it equipment, cabinets and room.�� as a result, resilience and efficiency are dramatically impacted by it equipment refresh and locations changes.

key concept: cpm is effective for maintaining resilience and energy efficiency as it configurations change over time

the simple system shown in figure 5 will be used to show how cooling paths #2 and #3 are defined by the interactions between thermal footprint and the environment and how cpm is used to improve cooling system efficiency.two versions of this system will be examined.

figure 5:� a simple data center system with one air handler and three different models of it equipment

the first has a typical raised floor configuration that supplies air to a floor tile in front of the cabinet within which sit the it equipment.this design does not constraint cooling path segments #2 and #3 as shown in figure 6.

figure 6:� cooling path segments #2 and #3 are completely unconstrained

further, air volume is oversupplied by 50%.oversupply of air and unconstrained cooling paths create significant amounts of bypass and air recirculation which significantly reduces cooling system efficiency.

figure 7 is an isolated view of the blade chassis server at the bottom of the cabinet to show cooing path segment #2 in detail for this unit.

figure 7:� isolated view of cooling path segment #2 for the blade chassis which shows significant mixing of supply and return air

here, 44% of the supply air comes from the air handler (dark blue) and 56% comes from the it equipment exhausts (light blue).the average intake temperature for this unit is 73 �f even though the supply air temperature from the air handler is 63 �f.cooling path segment #2 for this unit is only 44% efficient in terms of air supply.

the second design option for this facility has air ducts that completely constrain cooling paths #2 and #3 as shown in figure 8.

figure 8:� cooling path segments #2 and #3 are completely constrained

further, the air handler is supplying the exact amount of air in cfm that is required by the it equipment.as a result, bypass and recirculation are completely eliminated and the cooling paths are100% efficient.perfect constraint of these segments maximizes the cooling effect of the air handler, drops the average inlet air temperature by about 13 �f and reduces the cfm requirement by 50%.�� with the extra margin on inlet temperature, the chilled water set point can be increased by 13 �f.by matching the supply to the demand, fan speed can be reduced by 30%. this results in a 20% and a 50% reduction in power consumed by the chiller and fans respectively.table 1 summarizes the results of cpm on the efficiency of this simple data center system.

version

description

% bypass

% recirculation

average inlet temperature

cooling system power requirement

1

unconstrained cooling paths

45%

39%

75.6 �c

--

2

fully constrained cooling paths

0%

0%

62.1 �c

reduced by 46%

table 1: the effect of cooling path management on cooling system efficiency

summary

in real-world data centers, complete cooling path constraint is not practical as it configuration flexibility is often needed to meet business requirements.this makes cpm for real-world data centers an ongoing effort to minimize the negative impact of partially unconstrained cooling paths on resilience and efficiency.this is a difficult to impossible task without the benefit of a simulation-based methodology like cpm and a cfd tool that can reveal the full details of cooling path segments #2 and #3.finally, visibility of the entire cooling path makes cpm a useful and cost effective approach initial cooling system design and for ongoing change management of the facility to maintain acceptable levels of it equipment resilience and energy efficient operation over its lifespan.

references

for more information see, tschudi, greenburg, mills, et. al. (2006) best practices for data centers: lessons learned from benchmarking 22 data centers, aceee summer study on energy efficiency in buildings, berkeley, ca

for more information see, malone, c. and belady, c. (2006) metrics to characterize data center it equipment energy use. proceedings of 2006 digital power forum, richardson, tx

Choose category and click GO to search for thermal solutions

 
 

Subscribe to Qpedia

a subscription to qpedia monthly thermal magazine from the media partner advanced thermal solutions, inc. (ats)  will give you the most comprehensive and up-to-date source of information about the thermal management of electronics

subscribe

Submit Article

if you have a technical article, and would like it to be published on coolingzone
please send your article in word format to [email protected] or upload it here

Subscribe to coolingZONE

Submit Press Release

if you have a press release and would like it to be published on coolingzone please upload your pr  here

Member Login

Supplier's Directory

Search coolingZONE's Supplier Directory
GO
become a coolingzone supplier

list your company in the coolingzone supplier directory

suppliers log in

Media Partner, Qpedia

qpedia_158_120






Heat Transfer Calculators