ASHRAE OR-10-024-2010 Contamination in the Data Center-The Impact of CRAC Belt Dust《数据中心的污染 CRAC带上灰尘的影响》.pdf

1、2010 ASHRAE 223ABSTRACTThere has been some speculation within the Data CenterIndustry that a source of contamination that may negativelyimpact the life of the IT equipment may be the blower systembelts in the CRAH units. Rather than debate the impact “beltdust” may have on the IT equipment, this pap

2、er is a study ofthe actual design of the belts and sheaves to determine if thereis a potential to eliminate the wear on the belts, thus eliminat-ing the potential for belt dust.Defining the Scope of the StudyThere are three causes of excessive belt wear in aData Center application:1. The belt alignm

3、ent. Proper alignment has the belt exactly90 to the shafts being driven. Mis-alignment will causeside loading on the belt and imbalance between pairedbelts if more than one belt is used. Both of these result inexcessive belt heating thus causing the belt to becomebrittle and elongate.The alignment o

4、f the belt is impacted by the originalfactory design, but also by adjustments made in the fieldto adjust the blower speed. This is especially true if vari-able pitch pulleys are used. As the variable pitch pulley isadjusted the center distance between the pulleys ischanged as well as the center line

5、 of the pulley. That isbecause one side of the pulley is fixed in place while theother side is moved in and out to adjust how far the beltdrops into the pulley groove. If possible variable pitchpulleys should be changed to fixed dimension units oncethe desired dimension is known in the field.2. The

6、belt tension. As a system runs the belt heats up. Thiscan cause the belt to stretch and slip if the center distanceof the pulleys is not adjusted. Once a belt begins to slip,it heats up faster thus causing more slip and wear. Theheat also causes the belt to harden. This will result in beltcracking.3

7、. The actual belt designgeometry and material. This isthe main focus of this paper.Reviewing belt geometry and material, the population ofbelt designs used shows a significant variety especially whennon-OEM replacement belts are included. The non-OEM beltswere reported to have significantly shorter

8、belt life due toexcessive wear and stretching, thus were not part of this anal-ysis. There were reports of belt replacement every 3 monthswhen non-OEM belts were used. The belts selected by theOEMs were found to often be of special “high-end” specgrade design, often coming in matched pairs when two

9、beltswere to be used on the same system. This matching wasreported to insure exact common dimensions and characteris-tics to ensure balanced loading.For this analysis three of these spec grade deigns wereevaluated.1. A seamless design with milled belt edges.2. A “raw edge” design that is made from a

10、 neoprenecompound and has a lower cord line to increase flexibilityand extend life. The bottom of the belt has a heavy crackresistant fabric to reduce wear, extend flex life, andreduce dusting.3. A “wrapped” design that is made from a neoprenecompound with the basic B-section design. It is wrappedwi

11、th a dry clutching cover to reduce dusting. It has twoContamination in the Data Center The Impact of CRAC Belt DustJ. Fred Stack Dan FanninAssociate Member ASHRAEJ.F. Stack is vice president of marketing, Emerson Network Power/Liebert Solutions, St. Louis, MO. Dan Fannin is vice president of market-

12、ing, Emerson Industrial Automation.OR-10-024 2010, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2010, Vol. 116, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print

13、or digital form is not permitted without ASHRAEs prior written permission. 224 ASHRAE Transactionslayers of fabric crack barriers in the bottom of the beltto resist cracking and to extend service life.METHODThe test samples, equipment and methods used to conductthe test will be described in the foll

14、owing sections.Test SamplesThree types of v-belts were tested in this program. Thesewere: (1) Seamless milled, (2) Raw Edge, and (3) WrappedMolded. These belts will be referred to as Belts A, B, and Crespectively in this report. Twelve samples of each of the beltswere tested.The A belt is a seamless

15、 design and the belt edges aremilled. Belts are manufactured as matched sets for multiple-belt sheave drives.The B belt is made from the same fiber loaded neoprenecompound that is used in cog-type belts. Its cord line has beenlowered to increase flexibility and extend life. A heavy crackresistant fa

16、bric is used in the bottom of the belt to reduce wear,extend flex life, and reduce dusting. The belt angle is designedat 36 to better-fit the sheaves typically used in CRAC units.The corners are trimmed to reduce stress. It is manufactured toHVAC center distance variation tolerances to promotesmooth

17、er operation. Code 1 matching is standard.The C belt is made from a neoprene compound. It has thebasic B-section design. It is wrapped with a dry clutchingcover to reduce dusting. It has two layers of fabric crack barri-ers in the bottom of the belt to resist cracking and to extendservice life. It i

18、s molded to promote smoother operation. Code1 matching is standard.Test Setup and ProcedureEach test belt was installed on an SMT dynamometer. Theconfiguration of the drive is shown in Figure 4.Parallel misalignment was introduced in the test systemwith a 0.1-in. offset between the drive and driven

19、sheaves. Thisamount of parallel offset renders a 5 angular belt misalign-ment between the two sheaves when the center distance isfixed at 10.46 in. This is ten times the 1/2 of one degree mis-alignment recommended. This misalignment was intended torepresent a worst-case of operation for the belt. Th

20、e belt wastensioned to a 7 lbf(3.2 kgf) load with the deflection tester. Aninfrared, type K thermocouple was mounted on the test appa-ratus to capture the tight side belt temperature.At the beginning of the test the tensioned belt test systemwas started with an inverter to control acceleration and t

21、oprevent relative motion between the belt and sheaves. Whenfull speed was reached, the VFD was switched out and theFigure 1 Seamless A belt section view.Figure 2 B belt section view.Figure 3 C belt section view.Figure 4 Belt drive. 2010, American Society of Heating, Refrigerating and Air-Conditionin

22、g Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2010, Vol. 116, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission. ASHRAE Transactions 225system was supp

23、lied with line power. The test system operatedat no load for one hour.Next, full load was applied to the test system. Full loadwas 435 in-lbftorque (50 kgfm) at the driven sheave shaft. Atthe 290 rpm driven shaft speed, the rate of output work for thedrive system was about two HP.Test systems were o

24、perated at full load and speed for acontinuous 3000-h period. The following measurements andadjustments were made during testing:Belt tension was set according to the belt vendor catalogat the new belt tension setting throughout the test pro-gram.At the beginning of testing, belt tension was checked

25、after the first 24 h of operation.Belt tension was checked on weekly intervals after thefirst 24 h.Center distance was checked at the start of testing andrechecked whenever belt tensions were adjusted.Driver and driven shaft speeds were checked daily andrecorded.Tight side belt temperature was recor

26、ded at one-hourintervals during the testing. Temperature was measuredas the sample exited the driven sheave.RESULTSTest results are presented in Table 1.SUMMARY RESULTSThe Raw Edge B belt had 12% less change in CD, thanthe Seamless Belt A.The Raw Edge B belt had 76% less weight loss, than theSeamles

27、s Belt A.The Raw Edge B belt had 75% less change in hardness,than the Seamless Belt A.The Raw Edge B belt ran 5 cooler than the SeamlessBelt A.The Wrapped Molded C belt also performed better thanthe Seamless Belt A, but not as well as the Raw EdgeBelt B.CONCLUSIONBelt life and belt wear are dependan

28、t on several factorsfrom alignment to tension, but also belt design can have asignificant impact. Proper belt design selection can result ingreatly reduced belt wear reducing the replacement time inter-val up to five years.The reduced belt wear greatly minimizes the concern ofbelt dust negatively im

29、pacting the IT hardware.DISCUSSIONMichael Patterson, Intel: With the push to warmer ambienttemperatures, what will be the effect on belt life, contamina-tion, and performance? Dan Fannin: The push to higher ambient will not affect V-beltlife, contamination (wear), or performance. The acceptableopera

30、tions range for V-belts is 32F to 140F ambienttemperature.Temperature has a big influence on the life of V-belts,especially very hot or cold environments. There are things youcan do to enhance belt performance, in spite of temperatureextremes. Rubber Manufacturers Association TechnicalBulletin IP 3-

31、1 (1997) provides guidelines on “stock powertransmission belts manufactured by members of RMA.” Thefollowing are excerpts from this bulletin:Low Temperature“All stock power transmission belts manufactured bymembers of the RMA will maintain their flexibility to a temper-ature of 32F (36C). Below this

32、 temperature, the rubbercompounds used in these belts stiffen significantly and may crackwhen flexed. Compounding for less than 32F (36C) willreduce belt life at high temperatures. The torque required to turnthe drive and flex the belts also increases at low temperatures.Increased start-up torque ca

33、n be a problem on automotive engineaccessory drives in low temperature conditions due to batterydrain during cranking of a cold engine.”High Temperature“A power transmission belt may be considered heat resis-tant when it continues to give satisfactory service in a hightemperature environment or afte

34、r it has been exposed to hightemperatures. All stock power transmission belts manufacturedby members of the RMA are made of materials compounded forgood resistance to heat. Normally, belt drives operating in anambient temperature above 85F (30C) but below 140F(60C) can be handled adequately with a s

35、tock belt drive selec-Table 1. V-Belt Testing Summary of ResultsBelt Design1st 24 h Change in C.D.* in. (mm) Hrs. 25-3000 Change in C.D. in. (mm) Number of Times Belt Tension was Adjusted Belt Change in Mass (Grams) Belt Increase in Hardness (Shore A) Belt AverageOperatingTemperature F (C)A Seamless

36、 0.053 (1.35) 0.073 (1.85) 3 14.4 6.0 104 (40.0)B Raw Edge 0.053 (1.35) 0.058 (1.47) 3 3.5 1.5 99 (37.2)C Wrapped Molded 0.103 (2.62) 0.027 (0.59) 2 5.3 1.0 102 (38.9)Note: Data presented in this table is an average of the sample population.* C.D. = Center Distance 2010, American Society of Heating,

37、 Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2010, Vol. 116, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission. 226

38、ASHRAE Transactionstion, though some reduction in below life may occur. Drivesoperating in ambient temperatures above 140F (60C) shouldbe referred to a belt supplier for recommendations.”Don Sprinkle, Turner Broadcasting Systems, Atlanta,GA: How do you recognize “rough edge” belts if you wantedto us

39、e for a specific change-out?Dan Fannin: The rough edge, also known as a raw edge belt,does not have a fabric cover. The most common belts in themarket have two cosmetic looks. One type is commonly known as a classical belt orwrapped belt. The belt has a fabric wrap. This product has thecosmetic look

40、 of fabric weave on the sidewall of the belt. Thesidewall is the part of the belt that contacts the sheave.The second type is a raw edge belt. These belts have nofabric on the sidewall. Normally, these types of belts have afabric on the top of the belt and some designs have fabric onthe bottom of th

41、e belt. Of the raw edge belt products, there aretwo designs, notched and un-notched (cogged belts).According to the Department of Energy 2005 MotorSystems Tip Sheet #5 (www1.eere.energy.gov/industry/bestpractices/pdfs/replace_vbelts_motor_systemts5.pdf),“About one-third of the electric motors in the

42、 industrial andcommercial sectors use belt drives. Belt drives provide flex-ibility in the positioning of the motor relative to the load.Pulleys (sheaves) of varying diameters allow the speed of thedriven equipment to be increased or decreased. A properlydesigned belt transmission system provides hi

43、gh efficiency,decreases noise, requires no lubrication, and presents lowmaintenance requirements. However, certain types of beltsare more efficient than others, offering potential energy costsavings.The majority of belt drives use V-belts. V-belts use a trap-ezoidal cross section to create a wedging

44、 action on the pulleysto increase friction and improve the belts power transfer capa-bility. Joined or multiple belts are specified for heavy loads. V-belt drives can have a peak efficiency of 95% to 98% at thetime of installation. Efficiency is also dependent on pulleysize, driven torque, under- or

45、 over-belting, and V-belt designand construction. Efficiency deteriorates by as much as 5% (toa nominal efficiency of 93%) over time if slippage occurs,because the belt is not periodically retensioned.Cogged belts have slots that run perpendicular to the beltslength. The slots reduce the bending res

46、istance of the belt.Cogged belts can be used with the same pulleys as equivalentlyrated V-belts. They run cooler, last longer, and have an effi-ciency that is about 2% higher than that of standard V-belts.”As a result of the efficiency gain, we recommend the useof notched, raw edge belts.Robert Sull

47、ivan, Uptime Institute, Morgan Hill, CA: Fixedpitched shelves will eliminate the problem if alignment is heldtrue. Dan Fannin: The use of fixed pitch sheaves and proper align-ment are factors. Many factors can cause belt wear, but the beltdesign, materials, and construction are the most importantfac

48、tors. The design, materials, and construction of materials in aV-belt are the most important factors in the reduction of CRACbelt dust. The sheave sidewalls have a micro finish maximumof 120 mu. The micro finish is a factor on variable pitchsheaves or fixed pitch sheaves. This is a very smooth finis

49、h, butthis finish can cause wear to the belt surface. The belts recom-mended by the original manufacturer have the design, mate-rials, and construction characteristics to provide the longestlife and reduced belt wear in the application. The special is why we recommend following the originalmanufacturers recommendation for belt replacement.Fred Stack: I advised the audience that in computer room airconditioners, the belts are generally 99% of the time in thecooler supply air stream. Thus, the longer life. 2010, American Society of Heating, Refrigerating and